article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the cosmological constant problem @xcite-@xcite , the accelerated expansion of the late time universe@xcite , the cosmic coincidence @xcite are challenges for the foundations of modern physics ( see also reviews on dark energy@xcite,@xcite , dark matter @xcite and references therein ) .
numerous models have been proposed with the aim to find answer to these puzzles , for example : the quintessence@xcite , coupled quintessence@xcite , @xmath2-essence@xcite , variable mass particles@xcite , interacting quintessence@xcite , chaplygin gas@xcite , phantom field@xcite , tachyon matter cosmology@xcite , brane world scenarios@xcite , etc .. these puzzles have also motivated an interest in modifications and even radical revisions of the standard gravitational theory ( general relativity ( gr))@xcite,@xcite . although motivations for most of these models can be found in fundamental theories like for example in brane world@xcite , the questions concerning the einstein gr limit and relation to the regular particle physics , like standard model , still remain unclear .
the listed and a number of other cosmological and astrophysical problems generate fundamental questions which are directly addressed to particle physics , like : what is the dark matter from the point of view of particle physics ; what is the relation between dark energy and the vacuum of particle physics after gauge symmetry breaking ; how the latter question is related to the cosmological constant problem ; what is the precise physics of the inflation of the early universe , etc .. on the other side , even without the cosmological input , particle physics has its own fundamental questions , like for example : what is the origin of fermionic generations ; whether is it possible to realize spontaneous gauge symmetry breaking without using tachyonic mass term .
it is very hard to imagine that it is possible to propose ideas which are able to solve all the above mentioned problems keeping at the same time unchanged the basis of fundamental physics , i.e. gravity and particle field theory .
this paper may be regarded as _
an attempt to find satisfactory answers at least to a part of the fundamental questions on the basis of first principles _ ,
i.e. without using semi - empirical models . in this paper
we explore the so called two measures field theory ( tmt ) where gravity ( or more exactly , geometry ) and particle field theory intertwined in a very non trivial manner , but the einstein gr is restored when the fermion matter energy density is much larger than the vacuum energy density . here
we _ have no purpose _ of constructing a complete realistic cosmological scenario .
instead , we concentrate our attention on studying a number of _ new surprising enough effects _ of tmt . we study the model which possesses gauge and scale symmetry ( the latter includes the shift symmetry of the dilaton field @xmath0 ) .
the scale symmetry is spontaneously broken due to intrinsic features of the tmt dynamics . except for peculiar properties of the tmt action
, the latter does not contain any exotic terms , i.e. _ only terms presented in standard model considered in curved space - time are present in our action_. the scalar sector dynamics represents an explicit example of @xmath2-essence resulting from first principles . in context of cosmological dynamics
, we study in this paper only the scalar sector of the model which includes the dilaton @xmath0 and the higgs @xmath1 fields .
the higgs contribution to the action has no terms which could be treated as tachyonic mass terms .
the parameter space of the model is large enough that allows to find different regions where the following different effects can take place _ without fine tuning of the parameters and initial conditions _ :
\a ) possibility of a power law inflation driven by a scalar field @xmath0 which is followed by the late time evolution driven both by a small cosmological constant and a scalar field @xmath0 with a quintessence - like potential ; smallness of the cosmological constant can be achieved _ without fine tuning of dimensionfull parameters_. \b ) in another region of the parameters , there is a possibility of the power law inflation ended with damped oscillations of @xmath0 around the state with zero cosmological constant ( we want to emphasize again that this is realized without tuning of the parameters and initial conditions ) . thus this scenario includes _ a resolution of the old cosmological constant problem _ , at least at the classical level .
this becomes possible because one of the basic assumptions of the weinberg no - go theorem@xcite is violated in our model .
\d ) there is a region of the parameters where solutions for the late time universe possess the following unexpected features : the dark energy density @xmath10 increases asymptotically ( as @xmath6 ) approaching from below to a constant value ; the equation - of - state @xmath3 in the late time universe is @xmath4 and @xmath5 _ asymptotically approaches _ @xmath7 _ from below_. this effect is achieved without any exotic term in the action similar to what is present in phantom models@xcite .
\e ) in the model with both dilaton @xmath0 and higgs @xmath1 fields , the power law inflation consists of two stages : during the first stage @xmath1 remains close to its initial value ; about 60 e - folding before the end of inflation @xmath1 suffers a transition to the state of damping oscillations around @xmath11 . after the end of the power law inflation a transition of the higgs field to a broken gauge symmetry phase @xmath8 occurs via oscillations of @xmath1 and @xmath0 .
it is interesting that this early universe scenario is realized _ without tachyonic mass terms_. another remarkable feature is that _ the energy density in the symmetry broken phase is zero without tuning of the parameters and initial conditions_. since we have not yet studied the particle creation that must be due to oscillations of the higgs field , we are not able to present here a complete cosmological scenario including the described above effects in a consistent way .
nevertheless it is evident that the listed effects can be in principle be useful for a unified resolution of some of the cosmological and particle physics fundamental problems .
the organization of the paper is the following : in sec.ii we present a review of the basic ideas of tmt and formulate a gauge and scale invariant model .
the remaining part of the main text of the paper is devoted to study of the scalar sector dynamics of the model including dilaton @xmath0 and higgs @xmath1 fields .
however the way the @xmath1 field is introduced in our tmt model requires a special consideration of its properties in order to demonstrate that it is indeed able to perform all what one should demand from the higgs field , for example in the standard model .
this was the reason of the need to include also gauge fields and fermions into the description of the model although they are not the subject of this paper .
this is done in sec.ii and appendixes a and b. in appendix a we present equations of motion in the original frame .
appendix b is devoted to the spin - connections . using results of appendixes a and b , the complete set of equations of motion in the einstein frame
is given in sec.iic .
in sec.iii we give a detailed formulation of the scalar sector cosmological dynamics in the spatially flat frw universe . for pedagogical reasons
there is a need to present the results first without the higgs field : in secs.iv-vi for three different shapes of the effective potential we study the attractor behavior of the phase trajectories , the cosmological constant problem and possibility of super - acceleration .
main features of the cosmological dynamics driven both by the dilaton field and the higgs field are studied in sec.vii .
the discussion section gives a systematic analysis of the main ideas and results . in appendix c
we shortly review our previous results of tmt@xcite concerning the restoration of gr and resolution of the fifth force problem in normal particle physics conditions , i.e. when the local fermionic energy density is many orders of magnitude larger than the scalar dark energy density in the space - time region occupied by the fermion .
although this is not a subject of the present paper we believe that it is important for reader to know that in normal particle physics conditions tmt reproduces gr . in appendix
d we shortly discuss what kind of model one would obtain when choosing fine tuned couplings to measures @xmath12 and@xmath13 in the action .
some additional remarks concerning the relation between the structure of tmt action and the cosmological constant problem are given in appendix e.
tmt is a generally coordinate invariant theory where _ all the difference from the standard field theory in curved space - time consists only of the following three additional assumptions _ :
the first assumption is the hypothesis that the effective action at the energies below the planck scale has to be of the form@xcite-@xcite @xmath14 including two lagrangians @xmath15 and @xmath16 and two measures of integration @xmath13 and @xmath12 or , equivalently , two volume elements @xmath17 and @xmath18 respectively .
one is the usual measure of integration @xmath13 in the 4-dimensional space - time manifold equipped with the metric @xmath19 .
another is the new measure of integration @xmath12 in the same 4-dimensional space - time manifold .
the measure @xmath12 being a scalar density and a total derivative in string and brane theories see ref.@xcite .
] may be defined * either by means of four scalar fields @xmath20 ( @xmath21 ) , cf . with the approach by wilczek , @xmath22 * or by means of a totally antisymmetric three index field @xmath23 @xmath24 to provide parity conservation in the case given by eq.([phi ] )
one can choose for example one of @xmath20 s to be a pseudoscalar ; in the case given by eq.([aabg ] ) we must choose @xmath23 to have negative parity .
some ideas concerning the nature of the measure fields @xmath20 are discussed in ref.@xcite .
a special case of the structure ( [ s ] ) with definition ( [ aabg ] ) has been recently discussed in ref.@xcite in applications to supersymmetric theory and the cosmological constant problem .
it is assumed that the lagrangians @xmath15 and @xmath16 are functions of all matter fields , the dilaton field , the metric , the connection ( or spin - connection ) but not of the `` measure fields '' ( @xmath20 or @xmath23 ) . in such a case , i.e. when the measure fields enter in the theory only via the measure @xmath12 , the action ( [ s ] ) possesses an infinite dimensional symmetry . in the case given by eq.([phi ] ) these symmetry transformations have the form @xmath25 , where @xmath26 are arbitrary functions of @xmath27 ( see details in ref.@xcite ) ; in the case given by eq.([aabg ] ) they read @xmath28 where @xmath29 are four arbitrary functions of @xmath27 and @xmath30 is numerically the same as @xmath31 .
one can hope that this symmetry should prevent emergence of a measure fields dependence in @xmath15 and @xmath16 after quantum effects are taken into account .
important feature of tmt that is responsible for many interesting and desirable results of the field theory models studied so far@xcite-@xcite consists of the assumption that all fields , including also metric , connection ( or vierbein and spin - connection ) and the _ measure fields _ ( @xmath20 or @xmath23 ) are independent dynamical variables .
all the relations between them are results of equations of motion .
in particular , the independence of the metric and the connection means that we proceed in the first order formalism and the relation between connection and metric is not necessarily according to riemannian geometry .
we want to stress again that except for the listed three assumptions we do not make any changes as compared with principles of the standard field theory in curved space - time . in other words , all the freedom in constructing different models in the framework of tmt consists of the choice of the concrete matter content and the lagrangians @xmath15 and @xmath16 that is quite similar to the standard field theory .
since @xmath12 is a total derivative , a shift of @xmath27 by a constant , @xmath32 , has no effect on the equations of motion . similar shift of @xmath16 would lead to the change of the constant part of the lagrangian coupled to the volume element @xmath33 . in the standard
gr , this constant term is the cosmological constant .
however in tmt the relation between the constant term of @xmath16 and the physical cosmological constant is very non trivial ( see @xcite-@xcite ) . in the case of the definition of @xmath12 by means of eq.([phi ] ) , varying the measure fields @xmath20 , we get @xmath34 since @xmath35 it follows that if @xmath36 , @xmath37 where @xmath38 and @xmath39 is a constant of integration with the dimension of mass . in what follows we make the choice @xmath40 . in the case of the definition ( [ aabg ] ) , variation of @xmath23 yields @xmath41 that implies eq.([varphi ] ) without the condition
@xmath36 needed in the model with four scalar fields @xmath20 . one should notice _ the very important differences of tmt from scalar - tensor theories with nonminimal coupling _ : + a ) in general , the lagrangian density @xmath27 ( coupled to the measure @xmath12 ) may contain not only the scalar curvature term ( or more general gravity term ) but also all possible matter fields terms .
this means that tmt modifies in general both the gravitational sector and the matter sector ; b ) if the field @xmath12 were the fundamental ( non composite ) one then instead of ( [ varphi ] ) , the variation of @xmath12 would result in the equation @xmath42 and therefore the dimensionfull integration constant @xmath43 would not appear in the theory .
applying the palatini formalism in tmt one can show ( see for example @xcite and appendix c of the present paper ) that the resulting relation between metric and connection includes also the gradient of the ratio of the two measures @xmath44 which is a scalar field .
the gravity and matter field equations obtained by means of the first order formalism contain both @xmath45 and its gradient .
it turns out that at least at the classical level , the measure fields affect the theory only through the scalar field @xmath45 . the consistency condition of equations of motion has the form of a constraint which determines @xmath46 as a function of matter fields .
the surprising feature of the theory is that neither newton constant nor curvature appear in this constraint which means that the _ geometrical scalar field _
@xmath46 _ is determined by the matter fields configuration _ locally and straightforward ( that is without gravitational interaction ) . by an appropriate change of the dynamical variables which includes a conformal transformation of the metric
, one can formulate the theory in a riemannian ( or riemann - cartan ) space - time .
the corresponding conformal frame we call `` the einstein frame '' .
the big advantage of tmt is that in the very wide class of models , _ the gravity and all matter fields equations of motion take canonical gr form in the einstein frame_. all the novelty of tmt in the einstein frame as compared with the standard gr is revealed only in an unusual structure of the scalar fields effective potential ( produced in the einstein frame ) , masses of fermions and their interactions with scalar fields as well as in the unusual structure of fermion contributions to the energy - momentum tensor : all these quantities appear to be @xmath45 dependent .
this is why the scalar field @xmath46 determined by the constraint as a function of matter fields , has a key role in dynamics of tmt models .
the tmt models possessing a global scale invariance@xcite are of significant interest because they demonstrate the possibility of spontaneous breakdown of the scale symmetry .
in fact , if the action ( [ s ] ) is scale invariant then this classical field theory effect results from eq.([varphi ] ) , namely from solving the equation of motion ( [ varphib ] ) or ( [ adl1 ] ) . one of the interesting applications of the scale invariant tmt models@xcite is a possibility to generate the exponential potential for the scalar field @xmath0 by means of the mentioned spontaneous symmetry breaking even without introducing any potentials for @xmath0 in the lagrangians @xmath15 and @xmath16 of the action ( [ s ] ) . some cosmological applications of this effect have been also studied in ref.@xcite . in order to show that tmt is able to provide a realistic results for gravity and particle physics , we present here a model with the @xmath47 gauge structure as in the standard model ( with standard content of the bosonic sector : gauge vector fields @xmath48 and @xmath49 and higgs doublet @xmath50 ) .
although fermions ( as well as gauge bosons ) have no relation to the scalar sector dynamics studied in the present paper we include also fermions for two reasons : the first is to show that there exists a regime where the einstein gr and regular particle physics are realized simultaneously ( see appendix d ) ; the second reason is to show that the higgs field implements all what it does in the standard model .
we start from only one family of the so called `` primordial '' fermionic fields and @xmath51(exact definition of this term is explained in the last paragraph of the next subsection ) : the primordial electron @xmath52 and neutrino @xmath53 . just as in the standard @xmath47 gauge invariant model
, we will proceed with the following independent fermionic degrees of freedom : one primordial left lepton su(2 ) doublet @xmath54 and right primordial singlets @xmath55 and @xmath56 .
in addition , a dilaton field @xmath0 is needed in order to realize a spontaneously broken global scale invariance@xcite .
it governs the evolution of the universe on different stages : in the early universe @xmath0 plays the role of inflaton and in the late time universe it is transformed into a part of the dark energy .
according to the general prescriptions of tmt , we have to start from studying the self - consistent system of gravity and matter fields proceeding in the first order formalism . in the model including fermions in curved space - time , this means that the independent dynamical degrees of freedom are : all matter fields , vierbein @xmath57 , spin - connection @xmath58 and the measure @xmath12 degrees of freedom , i.e. @xmath20 or @xmath23 .
we postulate that in addition to @xmath47 gauge symmetry , the theory is invariant under the global scale transformations : @xmath59 if the definition ( [ aabg ] ) is used for the measure @xmath12 then the transformations of @xmath20 in ( [ stferm ] ) should be changed by @xmath60 .
this global scale invariance includes the shift symmetry@xcite of the dilaton @xmath0 and this is the main factor why it is important for cosmological applications of the theory@xcite .
we choose an action which , except for the modification of the general structure caused by the basic assumptions of tmt , _ does not contain any exotic terms and fields_. keeping the general structure ( [ s ] ) , it is convenient to represent the action in the following form : @xmath61 \nonumber\\ & & -\int d^{4}x e^{2\alpha\phi /m_{p}}[\phi v_{1}(h ) + \sqrt{-g}v_{2}(h ) ] + \int d^{4}x\sqrt{-g}l_{gauge } + \int d^{4}x e^{\alpha\phi /m_{p}}(\phi
+ k\sqrt{-g})l_{fk } \nonumber\\ & & -\int d^{4}xe^{\frac{3}{2}\alpha\phi /m_{p } } \left[(\phi + h_{e}\sqrt{-g})f_{e}\overline{l}_{l}\,h\,e_{r } + [ ( \phi + h_{n}\sqrt{-g})f_{n}\overline{l}_{l}\,h^{c}\,n_{r } + h.c.\right ] \label{totaction}\end{aligned}\ ] ] the notations in ( [ totaction ] ) are the following : @xmath62 ; the scalar curvature is @xmath63 where @xmath64 @xmath65 @xmath66 \label{lfkin}\ ] ] where @xmath67 @xmath68 and , as usual , @xmath69 for @xmath70 scalars and @xmath71 for @xmath70 spinors ; @xmath72 sets the standard hypercharges : @xmath73 , @xmath74 , @xmath75 ; @xmath76 and finally @xmath77 , @xmath78 . in ( [ totaction ] )
there are two types of the gravitational terms and of the `` kinetic - like terms '' ( both for the scalar fields and for the primordial fermionic ones ) which respect the scale invariance : the terms of the one type coupled to the measure @xmath12 and those of the other type coupled to the measure @xmath13 .
for the same reason there are two different sets of the yukawa - like coupling terms of the primordial fermions . using the freedom in normalization of the measure fields ( @xmath20 in the case of using eq.([phi ] ) or @xmath23 when using eq.([aabg ] ) )
, we set the coupling constant of the scalar curvature to the measure @xmath12 to be @xmath79 . normalizing all the fields such that their couplings to the measure @xmath12 have no additional factors , we are not able in general to provide the same in terms describing the appropriate couplings to the measure @xmath13 . this fact explains the need to introduce the dimensionless real parameters @xmath80 and we will only assume that _ they have close orders of magnitudes_. the real positive parameter @xmath81 is assumed to be of the order of unity ; in all solutions presented in this paper we set @xmath82 . for newton
constant we use @xmath83 , @xmath84 .
one should also point out the possibility of introducing two different potential - like exponential functions of the dilaton @xmath0 coupled to the measures @xmath12 and @xmath13 with factors @xmath85 and @xmath86 .
@xmath87 and @xmath88 must be @xmath0-independent to provide the scale symmetry ( [ stferm ] )
. however they can be higgs - dependent and then they play the role of the higgs pre - potentials ( we will see below how the effective potential of the scalar sector is generated in the einstein frame ) .
the choice of the action ( [ totaction ] ) needs a few additional explanations : \1 ) with the aim to simplify the analysis of the results of the model ( containing too many free parameters ) we have chosen the coefficient @xmath89 in front of @xmath13 in the first integral of ( [ totaction ] ) to be a common factor of the gravitational term @xmath90 and of the kinetic term for the higgs field @xmath50 . except for the simplification
there are no reasons for such a fine tuned choice .
\2 ) we choose the kinetic terms of the gauge bosons in the conformal invariant form which is possible only if these terms are coupled to the measure @xmath13 . introducing the coupling of these terms to the measure @xmath12 would lead to the nonlinear field strength dependence in the gauge fields equations of motion as well as to non positivity of the energy .
another consequence is a possibility of certain unorthodox effects , like space - time variations of the effective fine structure constant .
\3 ) one can show that for achieving the right chiral structure of the fermion sector in the einstein frame , one has to choose in ( [ totaction ] ) the coupling of the kinetic terms of all the left and right primordial fermions to the measures @xmath13 and @xmath12 to be universal .
this feature is displayed in the choice of the parameter @xmath2 to be the common factor in front of the corresponding kinetic terms for all primordial fermion degrees of freedom . except for these three items , eq.([totaction ] ) describes _ the most general action of tmt satisfying the formulated above symmetries_. in appendix a we present the equations of motion resulting from the action ( [ totaction ] ) when using the original set of variables . in appendix b one can find the equation and solution for the spin - connection .
the common feature of all the equations in the original frame is that the measure @xmath12 degrees of freedom appear only through dependence upon the scalar field @xmath45 , eq.([zeta ] ) . in particular , all the equations of motion and the solution for the spin - connection include terms proportional to @xmath91 , that makes space - time non riemannian and all equations of motion - noncanonical .
it turns out that with the set of the new variables ( @xmath0 , @xmath50 , @xmath49 and @xmath48 remain the same ) @xmath92 which we call the einstein frame , the spin - connections become those of the einstein - cartan space - time , see appendix c. since @xmath93 , @xmath94 and @xmath95 are invariant under the scale transformations ( [ stferm ] ) , spontaneous breaking of the scale symmetry ( by means of eq.(4 ) which for our model ( [ totaction ] ) takes the form ( [ app1 ] ) ) is reduced in the new variables to the _ spontaneous breakdown of the shift symmetry _
@xmath96 notice that the goldstone theorem generically is not applicable in this sector of the theory@xcite .
the reason is the following .
in fact , the shift symmetry ( [ phiconst ] ) leads to a conserved dilatation current @xmath97 . however , for example in the spatially flat frw universe the spatial components of the current @xmath98 behave as @xmath99 as @xmath100 . due to this anomalous behavior at infinity , there is a flux of the current leaking to infinity , which causes the non conservation of the dilatation charge .
the absence of the latter implies that one of the conditions necessary for the goldstone theorem is missing .
the non conservation of the dilatation charge is similar to the well known effect of instantons in qcd where singular behavior in the spatial infinity leads to the absence of the goldstone boson associated to the @xmath101 symmetry . after the change of variables to the einstein frame
( [ ctferm ] ) and some simple algebra , the gravitational equations ( [ app12 ] ) take the standard gr form @xmath102 where @xmath103 is the einstein tensor in the riemannian space - time with the metric @xmath104 ; the energy - momentum tensor @xmath105 is now @xmath106 the function @xmath107 has the form @xmath108 -v_{2}(\upsilon)}{(\zeta + b_{g})^{2 } } ; \label{veff1}\ ] ] @xmath109 is the canonical energy momentum tensor for the @xmath47 gauge fields sector ; @xmath110 is the canonical energy momentum tensor for ( primordial ) fermions @xmath94 and @xmath95 in curved space - time including also their standard @xmath47 gauge interactions .
@xmath111 is the _ noncanonical _ contribution of the fermions into the energy momentum tensor @xmath112 where @xmath113 and @xmath114 ( @xmath115 ) are respectively @xmath116 , \label{zeta}\ ] ] @xmath117 the structure of @xmath111 shows that it behaves as a sort of variable cosmological constant@xcite but in our case it is originated by fermions .
this is why we will refer to it as _ dynamical fermionic @xmath118 term_. this fact is displayed explicitly in eq.([tmn - noncan ] ) by defining @xmath119 .
one has to emphasize the substantial difference of the way @xmath119 emerges here as compared to the models of the condensate cosmology ( see for example refs.@xcite-@xcite ) . in those models the dynamical cosmological constant results from bosonic or fermionic condensates . in tmt model studied here , @xmath119 is originated by fermions but there is no need for any condensate . in appendix c
we show that @xmath120 becomes negligible in gravitational experiments with observable matter
. however it may be very important for some astrophysics and cosmology problems@xcite .
the dilaton @xmath0 field equation ( [ phi - orig ] ) in the new variables reads @xmath121 \nonumber\\ & -&\frac{\alpha}{m_{p}(\zeta + b_{g})^{2 } } \left[(\zeta + b_{g})m^{4}e^{-2\alpha\phi / m_{p}}-(\zeta -b_{g})v_{1}(\upsilon ) -2v_{2}(\upsilon)-\delta b_{g}(\zeta + b_{g})\frac{1}{2}\tilde{g}^{\alpha\beta}\phi_{,\alpha}\phi_{,\beta}\right ] \nonumber \\ & = & -\frac{\alpha}{m_{p}}[z_{n}(\zeta)m_{n}(\zeta , \upsilon)\overline{n^{\prime}}n^{\prime } + z_{e}(\zeta)m_{e}(\zeta,\upsilon)\overline{e^{\prime}}e^{\prime } ] \equiv -\frac{\alpha}{m_{p}}\lambda_{dyn}^{(ferm)}. \label{phief+ferm1}\end{aligned}\ ] ] the higgs field equation in the unitary gauge ( [ higgs - orig ] ) takes in the einstein frame the following form @xmath122 , \label{chief+ferm}\ ] ] where @xmath123 we have omitted here interactions to gauge fields which are of the canonical @xmath47 form due to eq.([dhiggs ] ) .
one can show that equations for the primordial leptons in terms of the variables ( [ ctferm ] ) take the standard form of fermionic equations for @xmath94 and @xmath95 in the einstein - cartan space - time where the spin - connection @xmath124 is determined by eq.([c7 ] ) and the standard gauge interactions are also present .
however the non - abelian structure of these interactions makes the resulting equations very bulky .
it is more convenient to represent the result of calculations for the lepton sector in the einstein frame in a form of the effective fermion action @xmath125 where @xmath126 - m_{n}(\zeta,\upsilon)\overline{n}^{\prime}n^{\prime } - m_{e}(\zeta,\upsilon)\overline{e}^{\prime}e^{\prime}. \label{lf - eff - ein}\ ] ] here @xmath127 and @xmath128 all the novelty in ( [ lf - eff - ein ] ) , as compared with the standard field theory approach to the @xmath47 unified gauge theory , consists of the @xmath45 dependence of the `` masses '' , eq.([muferm1 ] ) , of the primordial fermions @xmath94 , @xmath95 .
the scalar field @xmath45 in eqs.([tmn])-([lf - eff - ein ] ) is determined by means of the constraint ( [ app10 ] ) which in the new variables ( [ ctferm ] ) takes the form @xmath129 - 2v_{2}(\upsilon)-\delta b_{g}(\zeta + b_{g})\frac{1}{2}\tilde{g}^{\alpha\beta}\phi_{,\alpha}\phi_{,\beta}\right\ } \nonumber \\ & & = [ z_{n}(\zeta)m_{n}(\zeta)\overline{n^{\prime}}n^{\prime } + z_{e}(\zeta)m_{e}(\zeta)\overline{e^{\prime}}e^{\prime}]\quad \equiv \quad \lambda_{dyn}^{(ferm ) } , \label{constraint2}\end{aligned}\ ] ] where @xmath130 one should point out the interesting and very important fact : the same @xmath119 emerges in the following three different places : in the noncanonical contribution of the fermions into the energy momentum tensor ( [ tmn - noncan ] ) , in the effective coupling of the dilaton to fermions ( the right hand side of eq.([phief+ferm1 ] ) ) and in the constraint ( [ constraint2 ] ) .
notice the very important fact that as a result of our choice of the conformal invariant form for the kinetic terms of the gauge bosons in the action([totaction ] ) , the gauge fields do not enter into the constraint .
the gauge fields equations in the einstein frame become exactly the same as in the standard field theory approach to the @xmath47 unified gauge theory .
for example , eq.([gauge - orig ] ) for the gauge field @xmath49 in the fermionic vacuum reads now @xmath131 + \tilde{g}^{\mu\nu}\frac{1}{8}g^{\prime 2}\upsilon^2b_{\nu}=0 , \label{gauge - ein}\ ] ] and similar for @xmath48 with the appropriate non - abelian structure .
it is straightforward now to construct linear combinations of the gauge fields to produce the electromagnetic field @xmath132 and
@xmath133 and @xmath134 bosons . one can show that not only in the case of the @xmath47 gauge theory but also in more general gauge theories , like gut , gauge fields equations of motion in the einstein frame ( in tmtf ) coincide with the appropriate equations of the standard field theory approach to the gauge theory .
therefore after the higgs field develops a non zero vacuum expectation value ( vev ) , the higgs phenomenon takes place here exactly in the same manner as in the standard approach to the unified gauge theories : fermions and part of the gauge degrees of freedom become massive ( see eqs.([muferm1 ] ) , ( [ lf - eff - ein ] ) and ( [ gauge - ein ] ) ) .
hence , @xmath135 gauge model serves an illustrative example that in spite of the very specific general structure of the tmt action ( [ totaction ] ) , the scalar field @xmath50 ( or @xmath1 in the unitary gauge ) indeed plays the role of the higgs field .
however the detailed mechanism by means of which the symmetry breaking is implemented in tmt may be very much different from how it is done in the standard gauge theories .
in particular , we are going to show that it can be done without a tachyonic mass term in the action . applying constraint ( [ constraint2 ] ) to eq.([phief+ferm1 ] )
one can reduce the latter to the form @xmath136-\frac{2\alpha\zeta}{(\zeta + b_{g})^{2}m_{p}}m^{4}e^{-2\alpha\phi / m_{p } } = 0 , \label{phi - after - con}\ ] ] where @xmath45 is a solution of the constraint ( [ constraint2 ] ) . due to the constraint ( [ constraint2 ] ) which determines the scalar field @xmath45 as a function of scalar and fermion fields , generically fermions in tmt are very much different from what one
is used to in normal field theory .
for example the fermion mass can depend upon the fermion density . in appendix d
we show that if the local energy density of the fermion is many orders of magnitude larger than the vacuum energy density in the space - time region occupied by the fermion then the fermion can have a constant mass .
however this is exactly the case of atomic , nuclear and particle physics , including accelerator physics and high density objects of astrophysics .
this is why to such `` high density '' ( in comparison with the vacuum energy density ) phenomena we refer as `` _ normal particle physics conditions _ '' and the appropriate fermion states in tmt we call `` _ regular fermions _ '' . for generic fermion states in tmt
we use the term `` _ primordial fermions _ '' in order to distinguish them from regular fermions .
when neglecting the fermions and gauge fields , the origin of the gravity is the scalar sector of the matter fields which consists of the dilaton @xmath0 and higgs @xmath1 fields .
the gravitational , dilaton and higgs equations of motion in the einstein frame follow immediately from eqs.([gef])-([veff1 ] ) , ( [ phief+ferm1 ] ) and ( [ chief+ferm ] ) where one should ignore all fermionic and gauge fields terms ; the scalar @xmath45 is a solution of the constraint ( [ constraint2 ] ) which for finite @xmath45 has now the following form @xmath137 - 2v_{2}(\upsilon)-\delta b_{g}(\zeta + b_{g})x=0 , \label{constr - vac}\ ] ] where @xmath138 in the absence of fermions case , the scalar sector can be described as a perfect fluid with the following energy and pressure densities resulting from eqs.([tmn ] ) and ( [ veff1 ] ) after inserting the solution @xmath139 of the linear in @xmath45 eq.([constr - vac ] ) @xmath140 } \left[(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1})^{2}- 2\delta b_{g}(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1})x -3\delta^{2 } b_{g}^{2}x^2\right ] , \label{rho1}\ ] ] @xmath141 } \left[(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1})^{2}+ 2\delta b_{g}(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1})x + \delta^{2 } b_{g}^{2}x^2\right ] , \label{p1}\ ] ] where @xmath142 . in a spatially flat frw universe with the metric @xmath143 filled with the homogeneous scalar sector fields @xmath0 and @xmath1 , the dilaton @xmath0 and higgs @xmath1 field equations of motion take the form @xmath144 @xmath145 where @xmath50 is the hubble parameter and we have used the following notations @xmath146 @xmath147 @xmath148 @xmath149 } \nonumber \\ & \times & \left[(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1 } ) [ b_{g}(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1})-2v_{2 } ] + 2\delta b_{g}v_{2}x+3\delta^{2}b_{g}^{3}x^{2}\right ] \label{q3}\end{aligned}\ ] ] and @xmath150^{2 } } \nonumber \\ & & \times\left[\left(b_{g}(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1}-\delta b_{g}x ) -2v_{2}\right)v^{\prime}_{1}+(m^{4}e^{-2\alpha\phi / m_{p}}+v_{1}+\delta b_{g}x)v^{\prime}_{2}\right ] \label{q - h}\end{aligned}\ ] ] it is interesting that the non - linear @xmath151-dependence appears here in the framework of the fundamental theory without exotic terms in the original action ( [ totaction ] ) .
this effect results just from the fact that there are no reasons to choose the parameters @xmath89 and @xmath152 in the action ( [ totaction ] ) to be equal in general .
the above equations represent an explicit example of @xmath2-essence@xcite .
in fact , one can check that the system of equations ( [ gef ] ) , ( [ rho1])-([h1 ] ) ( accompanied with the functions ( [ q1])-([q - h ] ) and written in the metric @xmath143 ) can be obtained from the k - essence type effective action @xmath153 \label{k - eff},\ ] ] where @xmath154 is given by eq.([p1 ] ) .
in contrast to the simplified models studied in literature@xcite , it is impossible here to represent the lagrangian density of the scalar sector in a factorizable form .
for example even in the case @xmath155 , it is impossible to represent @xmath156 , eq.([p1 ] ) , in the form of the product @xmath157 .
recall that for the sake of simplicity we have chosen the coefficient @xmath89 in front of @xmath13 in the first integral of ( [ totaction ] ) to be a common factor of the gravitational term @xmath90 and of the kinetic term for the higgs field @xmath50 .
it is evident that without such a fine tuned choice we would obtain the k - essence type effective action non - linear both in @xmath151 and @xmath72 .
in this section we specialize the above model to the cosmological dynamics in a simplified version of tmt where the dilaton is the only matter field of the model .
the combined effect of both dilaton and higgs , leading to a new type of cosmological mechanism for the gauge symmetry breaking will be studied in sec.vii .
however , even in the toy model without the higgs field , the full analysis of the problem is complicated because of the large space of free parameters appearing in the action .
the qualitative analysis of equations is significantly simplified if @xmath158 .
this is what we will assume in this subsection .
although it looks like a fine tuning of the parameters ( i.e. @xmath159 ) , it allows us to understand qualitatively the basic features of the model .
in fact , only in the case @xmath158 the effective action ( [ k - eff ] ) takes the form of that of the scalar field without higher powers of derivatives .
role of @xmath160 in a possibility to produce an effect of superacceleration will be studied in sec.v .
so let us study the cosmology governed by the system of equations @xmath161 and ( [ rho1])-([phi1 ] ) where one should ignore all the higgs dynamics ( therefore @xmath87 and @xmath88 are now constants and @xmath162 ) and set @xmath158 . in the described approximation
the constraint ( [ constr - vac ] ) yields @xmath163 the energy density and pressure take then the canonical form , @xmath164 where the effective potential of the scalar field @xmath0 results from eq.([veff1 ] ) @xmath165^{2 } } { 4[b_{g}\left(v_{1}+m^{4}e^{-2\alpha\phi / m_{p}}\right)-v_{2 } ] } \label{veffvac - delta=0}\ ] ] and the @xmath0-equation ( [ phi1 ] ) is reduced to @xmath166 notice that @xmath167 is non - negative for any @xmath0 provided @xmath168 that we will assume in what follows .
we assume also that @xmath169 . in the models with @xmath170 :
fig.(a ) @xmath171 ( in the graph @xmath172 and @xmath173 ) ; fig.(b ) @xmath174 ( in the graph @xmath172 and @xmath175 ) .
the value of @xmath176 in the minimum @xmath177 is larger than zero ; fig.(c ) @xmath178 , @xmath179 ( in the graph @xmath180 and @xmath181 ) .
@xmath182 in the minimum @xmath183 . in all the cases here as well as in all solutions presented in this paper
we choose @xmath82.,width=680,height=453 ] in the following three sections we consider three different dilaton - gravity cosmological models determined by different choice of the parameters @xmath87 and @xmath88 in the action with @xmath158 : two models with @xmath170 and @xmath184 and one model with @xmath178 .
the appropriate three possible shapes of the effective potential @xmath167 are presented in fig.1 . a special case with the fine tuned condition @xmath185 is discussed in appendix d where we show that equality of the couplings to measures @xmath12 and @xmath13 in the action ( equality @xmath185 is one of the conditions for this to happen ) gives rise to a symmetric form of the effective potential .
in this model the effective potential ( [ veffvac - delta=0 ] ) is a monotonically decreasing function of @xmath0 ( see fig.[fig1]a ) . as @xmath186 ,
the effective potential ( [ veffvac - delta=0 ] ) behaves as the exponential potential @xmath187 .
so , as @xmath188 the model is able to describe a power law inflation of the early universe@xcite if @xmath189 . the latter condition will be assumed in all analytic solutions and qualitative discussions throughout the paper . applying this model to the cosmology of the late time universe and assuming that the scalar field @xmath190 as @xmath6 , it is convenient to represent the effective potential ( [ veffvac - delta=0 ] ) in the form @xmath191 where @xmath192 is the positive cosmological constant and @xmath193}. \label{v - quint - without - ferm - delta=0}\ ] ] we see that the evolution of the late time universe is governed both by the cosmological constant @xmath194 and by the quintessence - like potential @xmath195 .
( a ) ) , the energy density @xmath10 ( fig .
( b ) ) and the equation - of - state @xmath5 ( fig . ( c ) ) upon @xmath196 . here and in all the graphs of this paper describing scale factor @xmath197 dependences , @xmath198 is normalized such that at the end point of the described process @xmath199 .
the values of @xmath87 and @xmath88 are as in fig.1a .
the graphs correspond to the initial conditions @xmath200 , @xmath201 .
the early universe evolution is governed by the almost exponential potential ( see fig.1a ) providing the power low inflation ( @xmath202 interval in fig.(c ) ) .
after transition to the late time universe the scalar @xmath0 increases with the rate typical for a quintessence scenario . later on the cosmological constant @xmath203 becomes a dominated component of the dark energy that is displayed by the infinite region where @xmath204 in fig.(c).,width=680,height=226 ] versus @xmath0 ) for the model with @xmath171 .
all trajectories approach the attractor which in its turn asymptotically ( as @xmath190 ) takes the form of the straight line @xmath205.,width=529,height=302 ] the smallness of the observable cosmological constant , in this model given by @xmath194 , is known as the new cosmological constant problem@xcite .
there are two ways to provide the observable order of magnitude of the present day vacuum energy density by an appropriate choice of the parameters of the theory .
\(a ) if @xmath179 then there is no need for @xmath87 and @xmath88 to be small : it is enough that @xmath206 and latexmath:[$v_{1}/ mechanism@xcite,@xcite ) .
for instance , if @xmath87 is determined by the energy scale of electroweak symmetry breaking @xmath208 and @xmath88 is determined by the planck scale @xmath209 then @xmath210 .
the range of the possible scale of the dimensionless parameter @xmath89 remains very broad .
\(b ) if @xmath211 or alternatively @xmath212 and @xmath213 then @xmath214 .
hence the second possibility is to choose the _ dimensionless _
parameter @xmath169 to be a huge number . in this case
the order of magnitudes of @xmath87 and @xmath88 could be either as in the above case ( a ) or to be not too much different ( or even of the same order ) .
for example , if @xmath215 then for getting @xmath216 one should assume that @xmath217 . note that @xmath89 is the ratio of the coupling constants of the scalar curvature to the measures @xmath13 and @xmath12 in the fundamental action of the theory ( [ totaction ] ) .
taking into account our assumption that the dimensionless parameters @xmath218 , @xmath152 , @xmath2 and @xmath219 @xmath220 are of the close order of magnitude , their huge values can be treated as a sort of _ a correspondence principle _ in the tmt .
in fact , using the notations of the general form of the tmt action ( [ s ] ) in the case of the action ( [ totaction ] ) , one can conclude that if these dimensionless parameters have the order of magnitude @xmath221 then the relation between the `` usual '' ( i.e. entering in the action with the usual measure @xmath13 ) lagrangian @xmath222 and the new one @xmath223 ( entering in the action with the new measure @xmath12 ) is roughly speaking @xmath224
. it seems to be very interesting that such _
a correspondence principle may be responsible for the extreme smallness of the cosmological constant_. summing the above analysis we conclude that the effective potential ( [ veffvac - delta=0 ] ) provides a possibility for a cosmological scenario which starts with a power law inflation and ends with a small cosmological constant @xmath194 .
it is very important that the effective potential ( [ veffvac - delta=0 ] ) appears here as the result ( in a certain range of parameters ) of the tmt model undergoing the spontaneous breakdown of the global scale symmetry and @xmath179 was studied in ref.@xcite .
the application of the tmt model with explicitly broken global scale symmetry to the quintessential inflation scenario was discussed in ref@xcite . ] .
results of numerical solutions for such type of scenario are presented in figs.2 and 3 ( @xmath172 , @xmath173 ) the early universe evolution is governed by the almost exponential potential ( see fig.1a ) providing the power low inflation ( @xmath202 interval in fig.(c ) ) with the attractor behavior of the solutions , see ref.@xcite .
after transition to the late time universe the scalar @xmath0 increases with the rate typical for a quintessence scenario .
later on the cosmological constant @xmath203 becomes a dominated component of the dark energy that is displayed by the infinite region where @xmath204 in fig.(c ) .
the phase portrait in fig.3 shows that all the trajectories started with @xmath225 quickly approach the attractor which asymptotically ( as @xmath190 ) takes the form of the straight line @xmath205 .
qualitatively similar results are obtained also when @xmath87 is positive but @xmath88 is negative .
in this case the effective potential ( [ veffvac - delta=0 ] ) has the minimum ( see fig.1b ) @xmath228 for the choice of the parameters as in fig.1b , i.e. @xmath172 and @xmath175 , the minimum is located at @xmath177 .
the character of the phase portrait one can see in fig.[fig4 ] . , @xmath229 ) for the model with @xmath174 and @xmath184 ( the parameters are chosen here as in fig.1b ) .
trajectories started anywhere in the phase plane in a finite time end up at the same point @xmath230 which is a node sink .
however there exist two attractors ending up at @xmath231 , one from the left and other from the right in such a way that all phase trajectories starting with @xmath225 quickly approach these attractors.,width=604,height=302 ] and @xmath184 : typical dependence of @xmath0 ( fig.(a ) ) , the energy density @xmath10 ( fig.(b ) ) and the equation - of - state @xmath5 ( fig.(c ) ) upon @xmath196 where the scale factor @xmath198 normalized as in fig.2 .
the graphs correspond to the initial conditions @xmath232 , @xmath233 .
the early universe evolution is governed by an almost exponential potential ( see fig.1b ) providing the power low inflation ( @xmath202 interval in fig.(c ) ) . after arriving the minimum of the potential at @xmath234 ( see fig.1b and the point @xmath235 of the phase plane in fig.4 ) the scalar @xmath0 remains constant . at this stage the dynamics of the universe is governed by the constant energy density @xmath236 ( see the appropriate intervals @xmath237 in fig.(b ) and @xmath238 in fig.(c)).,width=680,height=226 ] for the early universe as @xmath239 , similar to what we have seen in the case of the monotonically decreasing potential in sec.iv , the model implies the power low inflation .
however , the phase portrait fig.[fig4 ] shows that now all solutions end up without oscillations at the minimum @xmath177 with @xmath240 . in this final state of the scalar field @xmath0 ,
the evolution of the universe is governed by the cosmological constant @xmath241 determined by eq.([minveff ] ) .
for some details of the cosmological dynamics see fig.[fig5 ] .
the desirable smallness of @xmath241 can be provided without fine tuning of the dimensionfull parameters by the way similar to what was done in item ( b ) of sec.iv .
the absence of appreciable oscillations in the minimum is explained by the following two reasons : a ) the non - zero friction at the minimum determined by the cosmological constant @xmath241 ; b ) the shape of the potential near to minimum is too flat .
the described properties of the model are evident enough after the shape of the effective potential ( [ veff1 ] ) in the einstein frame is obtained in tmt .
nevertheless we have presented them here because this model is a particular ( fine tuned ) case of an appropriate model with @xmath160 studied in the next subsection where we will demonstrate a possibility of states with @xmath4 without phantom in the original action .
the most remarkable features of the effective potential ( [ veffvac - delta=0 ] ) is that it is proportional to the square of @xmath275 . due to this , as @xmath178 and @xmath179 , _ the effective potential has a minimum where it equals zero automatically _ , without any further tuning of the parameters @xmath87 and @xmath88 ( see also fig.1c ) .
this occurs in the process of evolution of the field @xmath0 at the value of @xmath276 where @xmath277 this means that the universe evolves into the state with zero cosmological constant without tuning parameters of the model .
if such type of the structure for the scalar field potential in a usual ( non tmt ) model would be chosen `` by hand '' it would be a sort of fine tuning .
but in our tmt model it is not the starting point , _ it is rather a result _ obtained in the einstein frame of tmt models with spontaneously broken global scale symmetry including the shift symmetry @xmath278 .
later on we will see the same effect in more general models including also the higgs field as well as in models with @xmath160 .
note that the assumption of scale invariance is not necessary for the effect of appearance of the perfect square in the effective potential in the einstein frame and therefore for the described mechanism of disappearance of the cosmological constant , see refs.@xcite-@xcite and appendix e. on the first glance this effect contradicts the no - go weinberg theorem@xcite which states that there can not exist a field theory model where the cosmological constant is zero without fine tuning .
recall that one of the basic assumptions of this no - go theorem is that all fields in the vacuum must be constant .
however , this is not the case in tmt .
in fact , in the vacuum determined by eq.([veff=0 ] ) the scalar field @xmath279 is non zero , see eq.([zeta - without - ferm - delta=0 ] ) .
the latter is possible only if all the @xmath20 ( @xmath21 ) fields ( in the definition of @xmath12 by means of eq.([phi ] ) ) or the 3-index potential @xmath23 ( when using the definition of @xmath12 by means of eq.([aabg ] ) ) have non vanishing space - time gradients .
moreover , exactly in the vacuum @xmath280 the scalar field @xmath45 has a singularity .
however , in the conformal einstein frame all physical quantities are well defined and this singularity manifests itself only in the vanishing of the vacuum energy density .
we conclude therefore that the weinberg theorem@xcite is not applicable in the context of the tmt models studied here .
in fact , the possibility of such type of situation was suspected by s. weinberg in the footnote 8 of his review@xcite where he pointed out that when using a 3-index potential with non constant vacuum expectation value , his theorem does not apply .
the results of numerical solutions are evident enough , but we want to present them here because they will be useful for comparison with the results of the next section . for the potential of fig.1c ( where we have chosen @xmath180 and @xmath181 ) the results of numerical solutions are presented in figures 9,10,11 .
we return now to the more general models of the scalar sector ( see sec.iii ) where the parameter @xmath242 , defined by eq.([delta ] ) , is non zero .
however we still ignore here the higgs dynamics .
the appropriate simplified version of the cosmological dynamics is of interest to us because it allows , without non relevant complications , to demonstrate the possibility of solutions for the late time universe with equation - of - state @xmath4 ( super - acceleration ) , which are favored by the present data@xcite,@xcite , @xmath243 , @xmath172 and @xmath175 .
the phase plane is divided up into two dynamically disconnected regions by the line @xmath244 . to the left of this line @xmath245 and to the right @xmath246 .
the phase portrait in the left hand side , i.e. in the region @xmath245 , corresponds to processes similar to those of sec.v.a .
trajectories in the right hand side of phase portrait , i.e. in the region @xmath246 , correspond to processes with super - accelerating expansion of the universe.,width=642,height=453 ] ( fig.(a ) ) and of the energy density @xmath10 , defined by eq.([rho1 ] ) , ( fig(b ) ) in the regime of super - acceleration .
both graphs correspond to the initial conditions @xmath247 , @xmath248 ; @xmath10 increases approaching asymptotically @xmath249.,width=604,height=188 ] for the initial conditions @xmath247 , @xmath250.,width=302,height=188 ] so , we consider now the dynamics of the frw cosmology described by eqs.([cosm - phi ] ) , ( [ rho1 ] ) , ( [ phi1 ] ) , ( [ q1])-([q3 ] ) where @xmath87 and @xmath88 are constants and @xmath155 . adding the non zero @xmath242 to the parameter space
enlarges significantly the number of classes of qualitatively different models . among the new possibilities the most attractive one is the class of models giving rise to solutions for the late time universe with equation - of - state @xmath4 _ without resorting to negative kinetic term in the fundamental action _ as in the phantom field models@xcite . before choosing the appropriate parameters for numerical studies ,
let us start from the analysis of eq.([phi1 ] ) .
the interesting feature of this equation is that for certain range of the parameters , each of the factors @xmath251 ( @xmath252 ) can get to zero .
equation @xmath253 determines a line in the phase plane @xmath254 . in terms of a mechanical interpretation of eq.([phi1 ] ) , the change of the sign of @xmath255 can be treated as the change of the mass of `` the particle '' .
therefore one can think of situation where `` the particle '' climbs up in the potential with acceleration .
it turns out that when the scalar field is behaving in this way , the flat frw universe undergoes a super - acceleration .
there are a lot of sets of parameters providing this effect .
for example we are demonstrating here this effect with the following set of the parameters of the original action([totaction ] ) : @xmath82 , @xmath172 and @xmath175 used in sec.iv but now we choose @xmath243 .
the results of the numerical solution are presented in figs.6 - 8 .
the phase plane , fig.6 , is divided into two dynamically disconnected regions by the line @xmath256 . to the left of this line @xmath245 and to the right @xmath246 .
comparing carefully the phase portrait in the region @xmath245 with that in fig.[fig4 ] of the previous subsection , one can see an effect of @xmath160 on the shape of phase trajectories .
however the general structure of these two phase portraits is very similar . in particular
, they have the same node sink @xmath257 . at this point
`` the force '' equals zero since @xmath258 .
the value @xmath259 coincides with the position of the minimum of @xmath167 because in the limit @xmath260 the role of the terms proportional to @xmath242 is negligible . among trajectories converging to node @xmath231 there are also trajectories corresponding to a power low inflation of the early universe , which is just a generalization to the case @xmath261 of the similar result discussed in the previous subsection . on the right side of the phase plane fig.6 , i.e. in the region @xmath246 , all trajectories approach the attractor which in its turn asymptotically ( as @xmath190 ) takes the form of the straight line @xmath205 .
for a particular choice of the initial data @xmath247 , @xmath248 , the features of the solution of the equations of motion are presented in figs.7 and 8 .
the main features of the solution as we observe from the figures are the following : 1 ) @xmath0 slowly increases in time ; 2 ) the energy density @xmath10 slowly increases approaching the constant @xmath262 defined by the same formula as in eq.([lambda - without - ferm - delta=0 ] ) , see also fig.1b ; for the chosen parameters @xmath263 .
3 ) @xmath264 is less than @xmath7 and asymptotically approaches @xmath7 from below . qualitative understanding of the fact that the energy density @xmath10 approaches @xmath265 during the super - accelerated expansion of the universe is based on the shape of the effective potential , fig.1b , that would be in the model with @xmath158 .
however , the possibility of climbing up in the potential with acceleration can be understood only due to the effect of changing sign of @xmath255 , eq.([q1 ] ) , which becomes possible in the model with @xmath160 . due to such mechanism of the super - acceleration
it becomes clear why qualitatively the same behavior one observes for all initial conditions @xmath266 disposed in the region @xmath246 . versus @xmath0 ) for the model with @xmath178 and @xmath179 .
all trajectories started with @xmath225 quickly approach the attractor long before entering the oscillatory regime .
the region appropriate to the oscillatory regime is marked by point @xmath267 .
the oscillation spiral is not visible in fig.11 because of the choice of the scale along the axis y.,width=566,height=302 ] the power law inflation ends with damped oscillations of @xmath0 around @xmath268 determined by eq.([veff=0 ] ) . for the choice @xmath180 eq.([veff=0 ] )
gives @xmath269 .
( b ) the exit from the early inflation is accompanied with approaching zero of the energy density @xmath10 .
the graphs correspond to the evolution which starts from the initial values @xmath270 , @xmath271 .
, width=604,height=264 ] as function of the scale factor for the parameters and initial conditions as in fig.10 .
most of the time the expansion of the universe is a power law inflation with almost constant @xmath202 ; @xmath5 oscillates between @xmath7 and @xmath272 at the exit from inflation stage , i.e. as @xmath273 and @xmath274,title="fig:",width=321,height=302 ] as function of the scale factor for the parameters and initial conditions as in fig.10 .
most of the time the expansion of the universe is a power law inflation with almost constant @xmath202 ; @xmath5 oscillates between @xmath7 and @xmath272 at the exit from inflation stage , i.e. as @xmath273 and @xmath274,title="fig:",width=321,height=302 ]
turning now to the general case of the scalar sector , sec.iii , where both dilaton and higgs dynamics are taken into account , we have to make some nontrivial choices for the higgs pre - potentials @xmath281 and @xmath282 .
we will see that for achieving a gauge symmetry breakdown there is no need for tachyonic mass terms in @xmath281 and @xmath282 and even for selfinteraction terms .
we make the following simplest choice @xmath283 assuming @xmath284 in order to understand qualitatively what is the mechanism of the gauge symmetry breaking it is useful to start from the model with @xmath158 .
then from eqs.([veff1 ] ) and ( [ constr - vac ] ) we get for @xmath158 @xmath285^{2 } } { 4\left[\left(b_{g}v_{1}^{(0)}-v_{2}^{(0)}\right)+ \left(b_{g}\mu_{1}^{2}-\mu_{2}^{2}\right)\upsilon^{2}+b_{g}m^{4}e^{-2\alpha\phi / m_{p } } \right]}\label{veff+higgs}\ ] ] even in this simple model the effective potential of the scalar sector contains a very non - trivial coupling between dilaton , playing the role of inflaton , and higgs fields .
it is easy to see that this coupling disappears in the limit @xmath239 corresponding to the very early universe : @xmath286 therefore , in the very early universe , the minimization of @xmath287 is achieved at @xmath9 which means that the higgs field is in an unbroken symmetry phase without resorting to high temperature effects .
so , in the very early cosmological epoch , we have @xmath288 , @xmath289 and @xmath290 and therefore one can use the results of sec.vi in what it concerns to the early inflationary epoch .
it turns out , however , that this preliminary analysis is unable to give a complete scenario of the early inflationary epoch , see the next two subsections .
if we want @xmath291 to be positive definite and the symmetry broken state to be the absolute minimum we have to assume in addition that @xmath292 this will be our choice .
it is interesting that in the particular case @xmath293 the effective potential ( [ veff+higgs ] ) can be written in the form of the ginzburg - landau type potential for the higgs field @xmath1 @xmath294^{2 } , \label{v - bmu1=bmu2}\ ] ] where the coupling `` constant '' @xmath295 and the mass parameter @xmath296 are the following functions of @xmath0 : @xmath297^{-1 } ; \qquad \sigma^{2}(\phi)=\mu_{1}^{-2}\left(|v_{1}^{(0)}|-m^{4}e^{-2\alpha\phi / m_{p}}\right ) .
\label{lambda - bmu1=bmu2}\ ] ] if the higgs field were to remain in the unbroken phase @xmath11 , then according to sec.vi , the power law inflation would be ended with oscillations of the @xmath0-field accompanied with approaching zero of the energy density .
however decreasing @xmath298 causes that @xmath299 gets to be zero .
the continuation of this process changes the shape of the higgs dependence of the effective scalar sector potential such that @xmath9 turns into local maximum and a nonzero @xmath300 appears as the _ true minimum _ of the effective scalar field potential ( [ veff+higgs ] ) _ in the _ @xmath1 _
direction_. this happens as @xmath301 and @xmath280 satisfy the equation of the following curve @xmath302 which at the same time , _ is also the condition for the minimum in the _ @xmath0 _ direction _ ( see also eqs.([phi1])-([q - h ] ) as @xmath158 ) .
the remarkable feature of the effective potential ( [ veff+higgs ] ) is that in this minimum it _ equals zero without fine tuning of the parameters of the action_. this vacuum with zero cosmological constant is degenerate .
notice however that this degeneracy has no relation to any symmetry of the action .
this degeneracy is a particular manifestation of the well known more general feature of the systems with two interacting scalar fields : if both of them are dynamically important then as noted in ref.@xcite , there is no attractor behavior giving a unique route into the potential minimum , as it happens in the single field case .
however , numerical and analytic solutions show that in spite of this general statement , for the system under consideration , in a broad enough range of the parameters @xmath81 , @xmath242 , @xmath303 , @xmath304 , @xmath305 and @xmath306 , both fields , i.e @xmath0 and @xmath1 , are dynamically important but there exists _ an attractor behavior_. as a result of this there is only one very short segment in the line ( [ min - v ] ) where all the phase trajectories end in the process of the cosmological evolution . in other words ,
the magnitudes of @xmath307 and @xmath268 are determined by the values of the parameters of the model but dependence upon the initial values @xmath308 , @xmath309 , @xmath310 and @xmath311 is extremely weak . for the study of numerical and analytical solutions as well as for detailed qualitative analysis of all the stages of the scalar sector evolution during the cosmological expansion it will be sometimes convenient to work with the equations of motion ( [ phi1])-([q - h ] ) written in terms of dimensionless parameters and variables
. then eqs.([phi1 ] ) and ( [ h1 ] ) take the following form ( here we restrict ourselves with the choice the results are very similar : this is because the transition to @xmath312 phase is accompanied with @xmath260 . ]
@xmath158 ) @xmath313 @xmath314\tilde{\upsilon}=0 \label{h - dimensionless}\end{aligned}\ ] ] where eq.([cosm - phi ] ) and the following dimesionless parameters and variables have been used @xmath315 and @xmath316 where @xmath317 } , \label{veff - dimensionless}\ ] ] @xmath318 and @xmath151 is defined by eq.([x ] ) . in terms of the dimensionless quantities , the equation for the manifold of the true vacuum ( [ min - v ] )
takes the form @xmath319 it follows from the positivity of the r.h.s . of eq.([vac - min - dimensionless ] ) that @xmath320 vacuum expectation value @xmath321 is indeed a minimum of the effective potential ( [ veff - dimensionless ] ) in the @xmath322 direction if the frequency squared of the @xmath322 oscillations around the vacuum manifold ( [ vac - min - dimensionless ] ) is positive : @xmath323 we conclude from ( [ upper - bound - on - v ] ) and ( [ omega - dimensionless - positive ] ) that the vacuum expectation value @xmath324 has to satisfy the upper bound @xmath325 we are presenting here the results of the numerical solutions for equations of motion in the model with the following set of the parameters : @xmath326 the choice of the gut scale for the integration constant @xmath39 seems to us to be the most natural . but restricting with a single higgs field we proceed actually in a toy model .
, fig.(a ) , and the energy density @xmath10 , fig.(b ) , as functions of @xmath327 , typical for all sets of the initial conditions , eqs([5-in - cond ] ) . (
[ 6-th - in - cond ] ) . during a short time at the beginning @xmath328 because of the big value of the inflaton kinetic energy .
the graph of @xmath5 exhibit that most of the time of the evolution the equation - of - state is close to the constant @xmath202 .
this power low inflation ends ( i.e. @xmath329 . ) as @xmath330 . after the end of inflation the energy density @xmath10 quickly approaches zero .
the equation - of - state at this period is not well defined due to oscillations of the higgs field.,width=604,height=302 ] , fig.(a ) , and its time derivative , fig.(b ) , typical for all the initial conditions , eq.([5-in - cond ] ) ; here the solution for the set ( 4 ) of the initial conditions is displayed . during the first stage of the inflation
( the gsb inflation , see the main text ) @xmath1 varies very slowly from its initial value @xmath310 .
after the point @xmath331 , i.e. during the gsr inflation , @xmath1 is very close to zero ( oscillates around @xmath9 ) .
after the point @xmath332 the higgs - inflaton system fulfils a transition to `` zero cosmological constant and broken gauge symmetry '' phase where the higgs field oscillates with decaying amplitude around the vacuum manifold ( [ min - v]).,width=642,height=264 ] of the 4-dimensional phase trajectory corresponding to the solution for the set ( 2 ) of the initial conditions .
the function @xmath333 starts from @xmath334 and changes very slowly during the gsb inflation .
after transition to the gsr inflation ( after the point @xmath335 ) the phase trajectory approaches the attractor ( @xmath336 , @xmath9 , @xmath337 ) via oscillations of @xmath1 around @xmath9 , see below .
the last phase transition during of which the higgs - inflaton system evolves ( with damping oscillations of @xmath1 ) along the vacuum manifold ( [ vac - min - dimensionless ] ) is presented by the `` horn '' after the point @xmath338 .
the middle line of the horn is in fact the vacuum manifold ( [ vac - min - dimensionless]).,width=604,height=302 ] ; here the solution for the set ( 2 ) of the initial conditions has been used .
the vertical line @xmath339 exhibits the stage of the gsb inflation .
the large curve corresponds to the transition from the gsb inflation to the gsr inflation .
the point labeled by @xmath340 is the projection of the interval @xmath341 in fig.15 on the @xmath342 plane .
after @xmath340 the higgs field @xmath1 evolves via oscillations from the practically zero value to the gauge symmetry broken phase with the vacuum expectation value @xmath343 which in the graph is about 0.84 m .
, width=642,height=264 ] of five 4-dimensional phase trajectories corresponding to five sets of the initial conditions ( [ 5-in - cond ] ) , fig.(a ) .
similar to fig.11 in sec.vi , the trajectories exhibit quick approach to the attractor @xmath336 .
point @xmath331 corresponds to the beginning of the second stage of inflation ( the gsr inflation ) which succeeds the first stage of inflation ( the gsb inflation ) .
the phase trajectories become very close to the attractor @xmath336 long before the point @xmath331 .
point @xmath332 corresponds to the start of the phase transition to the `` zero cosmological constant and broken gauge symmetry '' phase .
fig.(b ) includes the scale factor dependence of the inflaton for five sets of the initial conditions ( [ 5-in - cond]).,width=642,height=264 ] of five 4-dimensional phase trajectories corresponding to five sets of the initial conditions ( [ 5-in - cond ] ) .
a surprising feature of the dynamics consists in the observed fact that independently of the initial conditions , all the phase trajectories achieve the line @xmath9 at the same value of @xmath0 , i.e. the point c is the common point for all phase trajectories .
moreover , it turns out that points f for different phase trajectories are also very close to each other : see fig.(b ) which zooms in on the the behavior of the five phase trajectories after the point f. two horns show that the five phase trajectories are divided into two directions giving @xmath343 of both signs ( recall that this sign has no physical meaning ) .
the final values of @xmath344 corresponding to different conditions are very close ( see however the appropriate discussion in the main text).,width=642,height=264 ] ) ) near and after the point @xmath335 in fig.18a .
solutions @xmath345 exhibit the behavior typical for bessel functions .
this feature as well as the reason of an observed synchronism will be explained in sec.viic.,width=566,height=302 ] around the line of the vacuum manifold quickly become practically equal for solutions with different initial conditions .
frequency of oscillations increases with @xmath1 , see eq.([omega - dimensionless - positive]).,width=566,height=302 ] in the main part of this subsection we choose @xmath158 .
but afterwards , at the end of the subsection , we show that the main results holds even if @xmath160 . to explore the effect of the initial conditions ( @xmath346 on the evolution of the inflaton @xmath0 and the higgs field @xmath1 we have solved the equations for six sets of the initial conditions .
figures 12 - 20 demonstrate the results for following five sets of them : @xmath347 the results for the sixth set of the initial conditions @xmath348 are shortly presented separately in fig.21 .
this is done just for technical reason because @xmath349 is too small .
so we are exploring the numerical solutions in the broad enough range of the initial values of the higgs field @xmath1 @xmath350 while according to ( [ upper - bound - on - v - full ] ) the true vacuum expectation value @xmath321 is bounded by @xmath351 fig.12 shows the behavior of the energy density and equation - of - state typical for all the initial conditions we have checked . for the set of the initial conditions ( 4 ) , eq.([5-in - cond ] ) , fig.13 shows the evolution of @xmath1 and @xmath352 as functions of the scale factor . using as a pattern the set of the initial conditions ( 2 ) , eq.([5-in - cond ] ) , in figs.14 - 16 we demonstrate the main features of the cosmological evolution of the higgs - inflaton system .
figs.17 - 20 allow to show that dependence of these features upon the initial conditions is very weak . in fig.21 we show that the interval of the initial conditions may be significantly expanded without altering the latter conclusion . in fig.22
we present the scale factor dependence of the higgs field for four sets of initial conditions different from those in eqs.([5-in - cond ] ) and ( [ 6-th - in - cond ] ) .
figs.23 and 24 show two additional facts : 1 ) the inflaton also oscillates during the transition to the symmetry broken phase , although its amplitudes are much less that those of the higgs field ; 2 ) exact detection of the finishing point @xmath353 of the pure classical transition to the symmetry broken phase is problematic . finally , in figs.25 - 26 we show that the effect of the parameter @xmath242 on the the main features of the cosmological evolution of the higgs - inflaton system , including the final gauge symmetry broken phase , is also very weak . ) hold for larger range of the initial values of @xmath1 . here for example we present the solution for the set ( [ 6-th - in - cond ] ) of the initial conditions where @xmath354 .
the locations of points c and f are the same as in fig.18b ; amplitudes of the oscillations of @xmath1 around the line of the vacuum manifold are very close to those in fig.20 .
the final value of @xmath344 is very close to those we have observed for the sets of the initial conditions ( [ 5-in - cond]).,title="fig:",width=321,height=302 ] ) hold for larger range of the initial values of @xmath1 . here for example we present the solution for the set ( [ 6-th - in - cond ] ) of the initial conditions where @xmath354 .
the locations of points c and f are the same as in fig.18b ; amplitudes of the oscillations of @xmath1 around the line of the vacuum manifold are very close to those in fig.20 .
the final value of @xmath344 is very close to those we have observed for the sets of the initial conditions ( [ 5-in - cond]).,title="fig:",width=321,height=302 ] exhibits that the gsr inflation starts about 50 e - folding before the end of inflation .
fig.(b ) zooms in on the interval near and after the point @xmath355.,width=680,height=302 ] : evolving together with @xmath1 along the vacuum manifold ( [ vac - min - dimensionless ] ) , @xmath0 performs a recoil with change of the sign of @xmath356 from positive to negative as the period average of @xmath356 gets to zero .
fig.(b ) shows details of the scalar factor dependence of @xmath0 at the end of the process displayed in fig.17b ( i.e. during the transition to the gauge symmetry broken phase ) for the same five sets of the initial conditions ( [ 5-in - cond]).,width=680,height=453 ] ) ) near to the region where the period average of @xmath356 gets to zero.,width=604,height=302 ] , eq.([delta ] ) , on the dynamics of the higgs - inflaton symbiosis .
four graphs in figs .
( a ) and ( b ) correspond to the solutions with the fixed initial conditions ( 3 ) in eq.([5-in - cond ] ) while for the parameter @xmath242 we have chosen the values @xmath158 , @xmath357 , @xmath358 , @xmath359 .
, width=680,height=302 ] -dependent ( as we see in fig.(a ) ) , the oscillatory regime of the approach of the higgs field to the final value of @xmath343 is practically independent of @xmath242.,title="fig:",width=302,height=302 ] -dependent ( as we see in fig.(a ) ) , the oscillatory regime of the approach of the higgs field to the final value of @xmath343 is practically independent of @xmath242.,title="fig:",width=340,height=302 ] fig.12b demonstrates the fact discussed in detail after eq.([min - v ] ) ) , that transition to the gauge symmetry broken phase at the same time is the transition to the state with zero vacuum energy density .
fig.12a demonstrates that most of the time of the evolution the equation - of - state is close to the constant @xmath360 .
this corresponds to a power low inflation which ends ( i.e. @xmath329 ) as @xmath330 and @xmath361 ( the latter one can see analyzing figs.12b and 17b ) .
the qualitative explanation of this effect is evident enough .
in fact , with our choice of the order of magnitudes of the parameters and initial conditions , contributions of @xmath362 strongly dominate over all the terms both in numerator and denominator of the potential ( [ veff - dimensionless ] ) as @xmath363 , i.e in terms of the dimensionless quantities([notations - dimensionless ] ) : @xmath364 \label{gg}\ ] ] therefore for @xmath363 the back reaction of the higgs field on the inflaton dynamics is negligible . therefore ignoring all what is related to the higgs field , the scalar sector potential @xmath287 , eq.([veff - dimensionless ] ) ,
acts as an exponential one depending only on the field @xmath365 ( see also eq.([veff+higgs - early ] ) ) , and our model coincides with the well studied power law inflation model@xcite . with our choice of @xmath82
the slow - roll parameter@xcite is a constant @xmath366 here are some other important features of the higgs - inflaton cosmological evolution we observe in the numerical solution : \1 . in spite of a practically constant equation - of state @xmath5 , fig.12a , it is convenient to distinguish between two stages of the inflation according to the value of the higgs field . starting from a non zero initial value
, @xmath1 varies very slowly during an initial stage of the inflation for two reasons : ( a ) a huge friction ( the hubble parameter @xmath367 , see the second term in eq.([h - dimensionless ] ) ) ; ( b ) the third term in eq.([h - dimensionless ] ) remains constant with very high accuracy as @xmath368 .
we will refer to this initial stage of the inflation as _ the gauge symmetry broken ( gsb ) inflation_. when the hubble parameter decreases significantly , the higgs field falls to its minimum @xmath9 or , more exactly it performs a transition to the phase where it oscillates with decaying amplitude around @xmath9 .
some features of this second stage of the inflation one can see in figs.13 , 14 , 15 , 17 , 18a , 19 , 21 and 22 .
we will refer to this stage of inflation as _ the gauge symmetry restored ( gsr ) inflation_. it lasts during the time corresponding to the interval which starts from the point @xmath335 and ends a little bit before the point @xmath338 .
projections of all phase trajectories on the plane @xmath369 ( see fig.17a ) approach very closely the attractor long before the end of inflation and even before transition to the gsr inflation .
it becomes clear now that the effect of the rapid approach to the attractor , we have observed in the simpler model of sec.vi , has _ a key role for the extremely weak dependence of the higgs field vacuum expectation value @xmath343 upon the initial values of @xmath308 and @xmath309_. \3 . in the process of the transition from the gsb inflation to the gsr inflation
, the higgs field gets to its first zero _ at the same value of _
_ for all the phase trajectories _ corresponding to different initial conditions , see point @xmath335 in figs.18a and 19 . the explanation of this surprising effect will be given in the next subsection on the the basis of an analytic solution . after point @xmath335
_ all the phase trajectories quickly approach the attractor _
( @xmath370 .
this approach occurs through the damped oscillations of @xmath1 around @xmath9 .
\4 . by means of graphs in figs.12a and 17b
one can check that with our choice of the parameters the inflation ends as @xmath330 and @xmath371 .
however it follows from figs.13a , 14a , 15 , 17b , 18 , 21 and 22a that the last phase transition starts as @xmath372 and @xmath373 .
therefore the _ transition to the `` zero cosmological constant and broken gauge symmetry '' phase starts after the end of inflation_. this result is also independent of the initial conditions .
the initial stage of the exit from inflation _ happens as @xmath374 and , as a result , the scalar sector energy density @xmath375 , eqs.([rho - dimensionless ] ) and ( [ veff - dimensionless ] ) , starts to go to zero ( see fig.12b ) .
the difference from the simple model of sec.vi here is that in the continuation of the exit from inflation , instead of entering the regime of oscillations of the inflaton , the modulus of the higgs field starts to increase in such a way that the higgs - inflaton system very fast approaches the vacuum manifold ( [ min - v ] ) , i.e. the higgs field @xmath1 appears in the new phase where @xmath376 .
afterwards the higgs - inflaton evolution proceeds along the vacuum manifold ( see the horn in fig.15 ) via oscillations of both the higgs field @xmath1 and the inflaton @xmath0 .
however there is an essential difference between the character of their oscillations : @xmath377 oscillates around zero while @xmath356 generically oscillates without changing its sign .
besides , the amplitude of the @xmath1-oscillations is a few orders of magnitude larger than that of the inflaton .
the character of the damped oscillations of @xmath1 one can see in figures.16 , 18b , 20 , 21 .
some details concerning the @xmath0 oscillations are presented in fig.24 .
the point @xmath378 of the vacuum manifold ( [ min - v ] ) where the higgs - inflaton system stops its classical evolution depends generically on the initial conditions . due to the attractor behavior of the phase trajectories discussed in items 2 and 3
, the initial conditions can affect the values of @xmath343 and @xmath379 only through the following two ways : a ) the amplitudes of the residual oscillations of @xmath1 around @xmath9 ( see fig.19 ) which are extremely small in the region close to point @xmath338 in fig.18a : up to @xmath380 the amplitudes of the classical oscillations decay about 30 orders of magnitude according to the behavior of the bessel function , see the next subsection and fig.28 therein ; b ) the residual velocity of the inflaton @xmath381 which is also very small in the region close to point @xmath338 . these two circumstances explain why for different initial conditions @xmath308 , @xmath309 , @xmath310 , @xmath311 there is only one short segment in the line ( [ min - v ] ) where all the phase trajectories end in the process of the cosmological evolution .
we have found that for our choice of the parameters ( [ parameters ] ) the value of @xmath382 approaches the interval @xmath383 .
\7 . analyzing the behavior of the numerical solution after the point @xmath338
we conclude that most of the higgs field kinetic energy accumulated in the @xmath322 oscillations around the mean value @xmath321 which monotonically evolves together with @xmath365 .
the kinetic energy of @xmath365 consists of two parts : the one due to the residual velocity of the monotonic evolution of @xmath365 and the other due to oscillations of @xmath384 , see fig.24 .
this circumstance causes a behavior anomalous on the first glance : after the period average of @xmath384 goes through zero , the higgs and inflaton fields start to go back , that one can see in figs.14a , 17b , 23 and 24 .
this recoil with the change of the sign of @xmath385 from positive to negative can happen in the process of the @xmath322-oscillations at the moment when @xmath386 appears to be negative ( see eq.([phi - dimensionless ] ) ) .
once @xmath365 starts to evolve backwards , this causes the same for @xmath343 . after some time
the period average of @xmath384 goes through zero again and the recoil with the change of the sign of @xmath385 from negative to positive happens now .
this process has apparently a tendency to repeat itself with decaying amplitude of the period average of @xmath384 , but attempts of more exact numerical solutions with the aim to obtain a definite final value of @xmath382 , where the evolution of our system finishes , run against a computational problem : enlarging running time of the process in our computations it is impossible ( or may be very hard ) to observe a certain point in the interval of vacuum manifold ( although short enough ) where the trajectories stop .
the main reason consists in the fact that not only the elastic forces in eqs.([phi - dimensionless ] ) and ( [ h - dimensionless ] ) approach zero ( due to approaching zero the amplitude of the oscillations around zero of the factor @xmath386 ) but also the friction caused by the cosmological expansion @xmath367 approaches zero .
this makes the classical mechanism of dissipation of the residual kinetic energies of @xmath365 and @xmath1 practically ineffective .
however the described problem should disappear after taking into account evident quantum effects , see sec.viie .
we have chosen @xmath387 to be the value of the scale factor at the end of the studied evolution process . in this section
@xmath387 is the value of the scale factor when the ( last ) transition to the `` zero cosmological constant and broken gauge symmetry '' phase practically ends . on the other hand ,
as we already discussed ( see the item 3 ) , after the point @xmath335 in fig.18a all the phase trajectories corresponding to different initial conditions appear to be very close to the attractor @xmath388 and this fact holds true till the last phase transition .
this explains why in the interval of @xmath327 from its value corresponding to the fall to the point @xmath335 up to @xmath389 , we observe practically the same pictures for all initial conditions ( see for example fig.22 ) .
in particular with our choice of the parameters , we have found that the transition from the gsb inflation to the gsr inflation ends about 50 e - folding before the end of inflation , and this result is practically independent of the initial conditions .
we have checked by means of numerical solutions that in models with @xmath160 all the above conclusions are not affected , see figs.25 and 26 .
this is clear enough because contributions of the terms with @xmath242 in equations of sec.iiia become negligible as the inflaton kinetic term is very small .
but this is exactly what happens due to the attractor behavior of the phase trajectories . to understand the mechanism responsible for the character of the attractor behavior of the phase trajectories ( observed in the numerical solution ) as they approach the point @xmath335 of the line @xmath390 in figs.18a and 19 , one should to take into account in eqs.([phi - dimensionless])-([x - dimensionless ] ) the inequalities ( [ gg ] ) which hold for @xmath391 . when doing this in a consistent way one can rewrite eqs.([phi - dimensionless ] ) and ( [ h - dimensionless ] ) , for @xmath392 , with high enough accuracy in the following simplified form @xmath393^{1/2}\frac{d\varphi}{d\tau } -\frac{\alpha}{2 } e^{-2\alpha\varphi } = 0 , \label{phi - dimensionless - simple}\ ] ] @xmath394^{1/2}\frac{d\tilde{\upsilon}}{d\tau } + \frac{1}{2}\left(\frac{m_p}{m}\right)^2(\tilde{\mu}_1 ^ 2+\tilde{\mu}_2 ^ 2)\tilde{\upsilon}=0 , \label{h - dimensionless - simple}\ ] ] while in the energy density @xmath375 in eq.([rho - dimensionless ] ) only the inflaton contribution has to be taken into account .
it is well known that eqs.([phi - dimensionless - simple ] ) and ( [ cosm - phi ] ) allow an exact solution corresponding to the power low inflation@xcite @xmath395 where @xmath396 and @xmath397 satisfy the condition @xmath398 note that in addition to the above restriction @xmath392 , the solution ( [ p - l - solution ] ) may be non applicable for a relatively short period of time from the very beginning where @xmath399 may be non monotonic , see fig.17a and 12a .
using the solution ( [ p - l - solution ] ) for @xmath400 one can rewrite eq.([h - dimensionless - simple ] ) in the form of the equation for @xmath401 : @xmath402 to obtain the analytic solution of this equation we use the following change of variable : @xmath403 . \label{change}\ ] ] then eq.([v(varphi ) ] ) takes the form @xmath404 general solution of eq.([v(z)-eq ] ) may be written in the form @xmath405 \label{v(z)-sol}\ ] ] where @xmath406 @xmath407 and @xmath408 are the bessel function of the first kind and the second kind respectively and @xmath409 , @xmath410 are two arbitrary constants . returning to the original variable @xmath365
, we obtain the general solution of eq.([v(varphi ) ] ) : @xmath411\cdot exp\left[-\left(\frac{3}{4\alpha}-\frac{\alpha}{2}\right)\varphi\right ] \label{v(phi)-sol-1}\ ] ] $ ] in eq.([v(phi)-sol-1 ] ) is very singular : to provide the initial value of @xmath412 to be of the order of one as @xmath413 ( fig.(a ) ) we are forced to choose the value of the constant @xmath414 .
then for example @xmath415 as @xmath416 ( fig.(b ) ) . , title="fig:",width=302,height=264 ] $ ] in eq.([v(phi)-sol-1 ] )
is very singular : to provide the initial value of @xmath412 to be of the order of one as @xmath413 ( fig.(a ) ) we are forced to choose the value of the constant @xmath414 .
then for example @xmath415 as @xmath416 ( fig.(b ) )
. , title="fig:",width=302,height=264 ] ) has been used in order to zoom in on one of the five patterns of the phase trajectories shown in figs.18a and 19 .
this graph illustrates a degree of damping of the higgs field oscillations as @xmath365 is close to -15 ( i.e. near to the end of the interval where the solution ( [ v(phi)-sol ] ) is applicable).,width=491,height=264 ] to compare this analytic solution with the results of the numeric solutions of subsection b we have used the same set of the values of the parameters ( [ parameters ] ) as in sec.b .
one can show that the graph of the first term proportional to @xmath417 has the shape which for five sets of the initial conditions ( [ 5-in - cond ] ) results in the graphs which coincide with those in fig.19 . but the graph of the second term proportional to @xmath418 exhibits a very singular behavior , see fig.27 .
therefore when trying to satisfy the initial conditions with natural , i.e. non anomalously large values of @xmath419 and @xmath420 , we obtain that @xmath421 is very close to zero . because of the extremely rapid decay of the second term , its contribution into the solution ( [ v(phi)-sol-1 ] ) quickly becomes negligible with growth of @xmath365 .
hence with very high accuracy one can proceed with the following analytic solution of eq.([v(varphi ) ] ) : @xmath422 \label{v(phi)-sol}\ ] ] choosing now the values of @xmath409 so that @xmath423 equals to the initial values considered in the numerical solutions in sec.viib , we obtain the same five graphs as in figs.18a and 19 ( but here only in the interval @xmath392 ) .
the curves @xmath424 , which are the projections of the 4-dimensional phase trajectories on the @xmath425-plane , have the following characteristic features : * from the beginning the solution @xmath426 remains almost constant in a long enough @xmath0-interval , that corresponds to the gsb stage of the inflation the analytic solution ( [ v(phi)-sol ] ) provides such a regime : the bessel function of the first kind @xmath417 starts from very small values and increases exponentially ; it turns out that the decaying exponential factor in eq.([v(phi)-sol ] ) compensates the growth of the bessel function @xmath417 with very high accuracy resulting in a very slow change of @xmath426 during an initial long enough stage of the power law inflation ( see fig.18a ) .
note that in contrast to this the bessel function of the second kind @xmath427 decreases exponentially and therefore its product with the decaying exponential factor in eq.([v(phi)-sol-1 ] ) makes its relative contribution to the solution negligible very quickly .
* analytic solutions @xmath426 with different initial conditions fall to the same point @xmath335 of the line @xmath428 , exactly as in the numerical solutions in sec.viib , figs.18a and 19 .
but now we have the mathematical explanation of this effect : the point @xmath335 is the first zero of the bessel function . since the choice of the initial value @xmath419 determines only the appropriate value of the factor @xmath409 in eq.([v(phi)-sol ] ) , it becomes clear why independently of initial values @xmath419 all the phase trajectories fall to the same point @xmath335 .
* after the point @xmath335 ( see figs.18a , 19 , 28 ) all the 4-dimensional phase trajectories approach the attractor @xmath429 whose projection on the @xmath430-plane we observe in figs.17a as the straight interval after the point @xmath431 . as we see in figs.19 , 28 , approaches to this attractor occur through the very strong damping of the oscillations of the higgs field around its zero value .
note also that locations of all the zeros of the bessel function @xmath417 are determined only by the parameters of the model and are independent of the initial conditions .
after the point @xmath432 , the presented analytic solution is not applicable since the inequality([gg ] ) does not hold for @xmath433 and therefore ignoring the back reaction of the higgs field on the inflaton dynamics does not hold anymore .
the fall of all the phase trajectories ( corresponding to different initial conditions ) to the same point @xmath335 of the attractor @xmath434 , along with the attractor behavior with respect to @xmath435 , has a decisive role in the observed effect : the magnitudes of @xmath301 and @xmath268 where @xmath1 and @xmath0 stop in the vacuum manifold ( [ min - v ] ) depend very weak of the initial conditions .
in fact , if different phase trajectories would fall to different points of the attractor @xmath434 then the degree of damping of their oscillations right before they enter into the regime of the last phase transition ( region @xmath338 in fig.18a ) could be generically very different . those of them whose deviations from the attractor were not small would have the limiting point @xmath436 of the phase transition significantly different from others .
the cosmological scenario studied in this section includes a number of dynamical effects demonstrating a strong enough tendency of the inflaton and higgs fields to coexistence and interference .
this is why we will refer to the appropriate phenomena as higgs - inflaton symbiosis ( his ) .
the dynamics of the his model resembles the hybrid inflation model@xcite but there are essential differences .
hybrid inflation models@xcite contain more than one scalar fields one of which ( the inflaton @xmath0 ) drives the early ( long ) stage of inflation and the dynamics of the other(s ) determines the character of the final ( very short ) stage of inflation .
one usually assumes@xcite that the extra scalar @xmath296 is the higgs field . due to a direct inflaton - higgs coupling the higgs
effective mass is @xmath0 dependent . during the early stage of inflation
the higgs effective mass square is positive and may be very large .
this is the reason why the higgs field is in the false vacuum @xmath437 during the early stage of inflation .
afterwards , slow roll of the inflaton below a certain value causes the change of the sign of the higgs effective mass square . in other words
, the inflaton plays the role of a trigger which causes transition of the higgs field into the true minimum of the scalar sector potential in the higgs direction .
this transition starts _ before _ the end of inflation because locations of the minima in the directions of the inflaton and the higgs field do not coincide .
the end of inflation and spontaneous breaking of the gauge symmetry appears to be intrinsically correlated : the last stages of inflation are supported not by the inflaton potential @xmath438 but by the `` noninflationary '' potential @xmath439 . to provide a desirable scenario
there is a need in strong enough restrictions on the parameters of the model .
note also that this type of models preserves the original feature of the standard approach to gauge symmetry breaking - the presence of a tachyonic mass term in the action .
here we summarize the main features of the his model including a number of differences from hybrid inflation models .
* incorporation of both the inflation and the gauge symmetry breaking is achieved _ without tachyonic mass _ terms in the original action ( see eqs.([totaction ] ) and ( [ v1v2higgs]),([m2>0 ] ) ) . * the inflaton - higgs coupling in eq.([veff+higgs ] ) disappears in the inflationary epoch , i.e as the inflaton @xmath239 , see eq.([veff+higgs - early ] ) .
therefore the _ effective higgs mass is constant during the inflation_. * the positivity of the effective mass square of the higgs field in the inflationary epoch ( see eq.([veff+higgs - early ] ) ) results directly from the positivity of the original mass square parameters @xmath440 and @xmath441 in the pre - potentials , eqs.([v1v2higgs]),([m2>0 ] ) .
* as a direct result of tmt , the effective scalar sector potential ( [ veff+higgs ] ) has a structure which provides automatically a _ common location _ of the minima in the inflaton and the higgs field directions .
these minima form the line ( [ min - v ] ) in the plane @xmath442 .
each point of this line describes a gauge symmetry broken state with zero vacuum energy density . instead of resorting to tachyonic mass terms ,
these minima are achieved by assigning negative values to the constant parts @xmath443 , @xmath444 of the pre - potentials , eqs.([v1v2higgs]),([m2>0 ] ) .
* constancy of the effective higgs mass during a significant part of the early stage of inflation allows for the higgs field @xmath1 to remain practically equal to its initial value when the hubble parameter is very large ( gsb inflation ) .
* the fall of the higgs field to the state @xmath9 ( gsr inflation ) occurs at a quite definite value of the e - folding before the end of inflation .
this value of the e - folding is practically independent of the initial conditions . * after falling to @xmath9 the higgs field performs quickly damping oscillations around @xmath9 .
* at the end of inflation ( i.e. when @xmath445 ) all the phase trajectories corresponding to different initial conditions appear to be very close to the attractor @xmath446 .
* the effective mass square of the higgs field _ becomes negative after the end of inflation_. exponential form of the @xmath0 dependence in eq.([veff+higgs ] ) makes this trigger effect very sharp .
the demonstrated extreme closeness of all phase trajectories ( corresponding to different initial conditions ) to the attractor provides that the phase transition to the gauge symmetry broken phase occurs into a very short interval of the values of @xmath343 ( and appropriate values of @xmath379 ) in the vacuum manifold([min - v ] ) .
* the remarkable structure of the scalar sector effective potential ( [ veff+higgs ] ) ( generated in the einstein frame in our tmt model ) automatically provides that the effective potential in the vacuum manifold ( [ min - v ] ) is equal to zero without fine tuning of the parameters of the action and the initial conditions .
* note finally that in contrast with the hybrid inflation models where strong enough restrictions on the parameters are needed , in the his model qualitatively the same results are obtained in a very broad range of the parameters satisfying the conditions ( [ m2>0 ] ) , ( [ bmu1>bmu2 ] ) . in the previous sections we did not touch quantum effects and their possible role in the studied processes .
one of the most important quantum effects is the particle creation .
here we want to discuss on the qualitative level what kind of influence one can expect from the particle creation on the studied processes . in our qualitative discussion
we have to take into account that in the his evolution there are _ two very different stages where oscillations of the higgs field must cause matter creation_.
\(1 ) _ the first oscillatory stage _
consists of the transition from gsb to gsr inflation and the gsr inflation itself .
a rapid transition of the higgs field from a nonzero value to the @xmath9 phase has to be accompanied with particle creation .
the subsequent coherent damping oscillations of the higgs field around @xmath9 , studied in detail in secs.viib and viic must be responsible for creation of massless particles , like for example _ massless gauge bosons_. this dissipative process results in an additional damping of the higgs field oscillations .
hence the quantum effect of particle creation acts here as an additional factor which forces the phase trajectories corresponding to different initial conditions to achieve closer unification with the attractor @xmath447 . as a result of this the final values of @xmath343 and @xmath379 in the symmetry broken phase
become more strongly independent of the initial conditions than in the pure classical model discussed in the preceding subsections .
one should note that with the chosen parameters , the reheating caused by the first stage of the particle creation can not play an essential role in the thermal history of the universe because it happens about 50 e - folding before the end of inflation .
however possible effect of the appropriate thermal fluctuations on the primordial density perturbations perhaps deserves a special study .
\(2 ) _ the second stage of the particle creation _ is realized after the end of inflation during the transition from @xmath434 phase to the gauge symmetry broken phase .
coherent oscillations of the higgs field around the vacuum manifold discussed in detail in sec .
viib may serve as a very effective mechanism of the preheating , for example via gauge bosons creation .
this effect has to be studied in detail , but it is beyond the goals of this paper .
nevertheless it is interesting to point out that _ the back reaction of the particle creation _ on the finish of the phase transition may be very important . indeed , as we discussed in the preceding subsections , the characteristic feature of the final stage of the classical evolution of the higgs - inflaton system is that the scalar sector energy density @xmath10 approaches zero .
if there is no other matter , the friction @xmath448 caused by the cosmological expansion also approaches zero and therefore the finish of the classical evolution of the higgs - inflaton system becomes problematic ( see item 8 in sec.viib ) .
however , if one takes into account the quantum effects of matter creation due to the higgs field oscillations around the vacuum manifold , this immediately produces additional friction in the higgs field equation non related to the cosmological expansion . besides , the presence of massive gauge bosons , photons and fermions in addition to the scalar sector ( the energy density of the latter is very close to zero )
means that the universe has entered into a radiation / matter dominated era .
therefore the friction caused by the cosmological expansion ( both in the inflaton and in the higgs field equations ) is now @xmath449 where @xmath450 includes also the radiation / matter energy density . due to these two back reaction effects the higgs and inflaton fields
must quickly dissipate their kinetic energy and the evolution of the higgs - inflaton system has to finish in a certain point of the vacuum manifold which is practically independent of the initial conditions .
one should finally point out that in this paper we have restricted ourselves only with a homogeneous distribution of the initial values of the higgs field .
however the discussed practical independence of @xmath343 of the initial conditions in the framework of the homogeneous problem may serve as an indication that in a more realistic case of inhomogeneous distribution of the initial values of the higgs field , its value in the final gauge symmetry broken phase will be practically the same everywhere in our universe .
the main idea of tmt is that the general form of the action @xmath451 is not enough in order to account for some of the fundamental problems of particle physics and cosmology .
the key difference of tmt from the standard field theory in curved space - time consists in the hypothesis@xcite-@xcite that in addition to the term in the action with the volume element @xmath18 there should be one more term where the volume element is metric independent but rather it is determined either by four ( in the 4-dimensional space - time ) scalar fields @xmath20 or by a three index potential @xmath23 , see eqs.([s])-([aabg ] ) .
we would like to emphasize that including in the action of tmt the coupling of the lagrangian density @xmath27 with the measure @xmath12 , we modify in general both the gravitational and matter sectors as compared with the standard field theory in curved space - time . besides we made two more assumptions :
the measure fields ( @xmath20 or @xmath23 ) appear only in the volume element ; one should proceed in the first order formalism . _
these assumptions constitute all the modifications of the general structure of the theory we have made as compared with the standard field theory in curved space - time_. in fact , the lagrangian densities @xmath27 and @xmath16 in the models we have studied in the present paper , contain all types of terms which should be present if the standard model would be formulated in curved space - time . however we do not consider any exotic terms and it turns out that there is _ no need for any exotic term _ in order to achieve desirable results .
in particular there is no need for the phantom type terms in @xmath27 and @xmath16 in order to obtain super - acceleration at the late time universe ; there is no need for tachyonic mass terms in the higgs pre - potentials @xmath87 and @xmath88 in order to obtain spontaneous breakdown of gauge symmetry . after making use of the variational principle and formulating the resulting equations in the einstein frame ,
we have seen that in the absence of fermions , the dynamics of the scalar sector is described by the effective action ( [ k - eff ] ) which is a concrete realization of the @xmath2-essence@xcite obtained from first principles . for pedagogical reasons in secs.iv-vi
we have studied first models without higgs field . as @xmath158 , the dynamics of @xmath0 can be analyzed by means of its effective potential ( [ veffvac - delta=0 ] ) . as @xmath239 the effective @xmath0 potential has the exponential form .
we have seen that independently of the values of the parameters @xmath87 , @xmath88 and under very general initial conditions , solutions rapidly approach a regime characterized by a power law inflation . if @xmath261 , we deal with the intrinsically @xmath2-essence dynamics .
the numerical solutions in this case have showed that there is no qualitative difference of the power law inflation from the case with @xmath158 .
for the model with the higgs field and @xmath158 , we have seen that with the choice of positive mass squared parameters @xmath440 and @xmath441 , the effective potential of the scalar sector in the very early universe , eq.([veff+higgs - early ] ) , describes the higgs field with positive mass square , @xmath441 would result in the tachyonic mass term in the effective potential ( [ veff+higgs - early ] ) in the early universe ; in such a case the minimization of the effective potential in the early inflation epoch could not be realized in the phase @xmath9 .
] , i.e. the minimization of the effective potential ( [ veff+higgs - early ] ) in the higgs direction in the inflationary epoch is achieved if the higgs field is in the unbroken symmetry phase @xmath9 .
the inflationary epoch with @xmath9 we called the gsr inflation .
however in contrast with the hybrid inflation@xcite , before the gsr inflation the universe suffers a long lasting gsb inflation , where the higgs field is in a broken gauge symmetry phase with slow varying @xmath452 . with our choice of the parameters , the transition from the gsb inflation to the gsr inflation occurs about 60 e - folding before the end of inflation , see fig.22 . in these toy models
there are three regions of the parameters @xmath87 and @xmath88 and appropriate three shapes of the effective potentials , fig.1
. therefore three different types of scenarios for exit from inflation can be realized : \a ) @xmath184 and @xmath171 , sec.iv . in this case
the power law inflation monotonically transforms to the late time inflation asymptotically governed by the cosmological constant @xmath194 .
\b ) @xmath226 , sec.v . in this case
the power law inflation is ended without oscillations at the final value @xmath453 , corresponding to the ( non zero ) minimum of the effective potential , .
\c ) @xmath178 and @xmath179 , sec.vi . in this case
the power law inflation is ended with damped oscillations of @xmath0 approaching the point of the phase plane ( @xmath454 ) where the vacuum energy @xmath455 .
this occurs without fine tuning of the parameters @xmath87 , @xmath88 , @xmath242 and the initial conditions . .
from the summary of the main features of the his ( see sec .
viid ) one can conclude that our tmt model implies a _ new cosmological mechanism for a transition to a gauge symmetry broken phase soon after the end of inflation_. from the field theory point of view ,
one of the advantages of the model is that the gauge symmetry breaking is obtained _ without tachyonic mass terms _ in the original action .
another important advantage of the model is that the inflaton driven transition to the the gauge symmetry broken phase is automatically ( i.e. without tuning ) accompanied with approaching zero of the scalar sector energy density . from the point of view of the realistic cosmology , the scenario based on the his model exhibits a new mechanism of reheating which most likely may be very effective : coherent oscillations of the higgs field must result in intensive creation of the gauge bosons .
besides , transition to the gauge symmetry broken phase starts soon after the end of the gsr inflation where @xmath456 and therefore masses of quantum of all the gauge fields are very close to zero at the beginning of the reheating . _ the old cosmological constant problem_. in secs.vi and vii , we have seen in details that if @xmath457 then , for a broad range of other parameters , the vacuum energy turns out to be zero without fine tuning ; this is possible both in the model with @xmath0 as the scalar sector and in the model with @xmath0 and @xmath1 .
this effect is a direct consequence of the result of the tmt : the effective scalar sector potential generated in the einstein frame is proportional to _ a perfect square_. if such type of the structure for the scalar field potential in a usual ( non tmt ) model would be chosen `` by hand '' it would be a sort of fine tuning . there is a need to explain how this result avoids the well known no - go theorem by weinberg@xcite stating that generically in field theory one can not have zero value of the potential in the minimum without fine tuning .
the reason why this theorem is not applicable here is that one of the basic assumptions of the theorem does not hold . indeed in the mentioned theorem
, it is assumed that all fields in the vacuum are constants . in our case ,
the basic assumption that the measure of integration in the action @xmath36 implies that all the measure fields ( @xmath20 ( @xmath21 ) in the definition ( [ phi ] ) or @xmath23 in the definition ( [ aabg ] ) ) have non vanishing space - time gradients .
in this paper we have restricted ourselves with the simplest form ( [ v1v2higgs ] ) of the higgs field dependence of the pre - potentials @xmath87 and @xmath88 .
if however we had studied models with more complicated higgs field dependence in @xmath87 and @xmath88 , including for example the quartic higgs self - interactions , then there could be vacua with zero cosmological constant disconnected from each other ( again without fine tuning ) .
this is an explicit realization of the `` multiple point principle '' proposal@xcite which is based on the idea that if there is a mechanism that sets a certain state to have a zero cosmological constant then the same mechanism may act also in other field configurations with the same result . .
interesting result following from the general structure of the scale invariant tmt model with @xmath458 is that the cosmological constant @xmath194 , eq.([lambda - without - ferm - delta=0 ] ) , is a ratio of quantities constructed from pre - potentials @xmath87 , @xmath88 and the dimensionless parameter @xmath89 .
such structure of @xmath194 allows to propose at least two independent ways ( see sec.iv ) for resolution of the problem of the smallness of @xmath194 that should be @xmath216 .
the first way is a kind of a _ seesaw _ mechanism@xcite .
for instance , if @xmath215 and @xmath459 then @xmath216 . the second way is realized if the dimensionless parameters @xmath89 , @xmath460 , etc . of the action ( [ totaction ] ) are huge numbers .
for example , if @xmath215 then for getting @xmath216 one should assume that @xmath217 .
it is interesting that this idea may be treated as an indication that there is a connection between the resolution of the new cosmological constant problem and a new type of _ the correspondence principle _ in the action , see details in item ( b ) of sec.iv .
we have seen in this paper that even restricting ourselves with only the scalar sector we discover a number of interesting effects which may be useful in the construction of a realistic scenario of the evolution of the universe . studying these effects makes it difficult in the same paper to address also questions like particle creation , perturbations , etc .. for this reason we can not yet present a more complete cosmological scenario . in particular ,
we have seen in sec.v that there is a region of the parameters where the late time universe can evolve with equation - of - state @xmath4 without introducing explicit phantom terms in the action .
this occurs due to the very interesting effect of the classical field dynamics described in sec.v in detail .
however it is impossible to obtain _ a pure classical solution _ which connects the early universe inflation with the late time super - acceleration .
this problem may be related with the toy character of the scenario where the role of the matter has been ignored . in tmt
, the effects related to particle creation can change the dynamics of the scalar sector .
in fact , if the particle creation results in the matter domination epoch , then in the presence of fermions , the constraint ( [ constraint2 ] ) differs from the constraint in the absence of fermions , eq.([constr - vac ] ) . therefore in the stage of a transition from the scalar sector ( dark energy ) domination epoch to the matter domination epoch
, @xmath45 must significantly vary that will lead to a change in the dynamics of the scalar sector due to the @xmath45-dependence of the scalar sector potential ( [ veff1 ] ) .
similar conclusion is true also for the dynamics of the scalar sector in the matter domination epoch itself .
in such a case the matter creation and subsequent matter domination epoch might be `` bridges '' connecting the early inflation with the possible late time super - acceleration .
this should include also a so far unclear mechanism responsible for the freezing of the higgs vacuum expectation value @xmath343 to survive till the late time universe .
this kind of questions we are going to study in the near future . in this paper
we have studied only classical two measures field theory and its effects in the context of cosmology .
however quantization of tmt as well as influence of quantum effects on the processes explored in this paper may have a crucial role .
we summarize here some ideas and speculations which gives us a hope that quantum effects can keep and even strengthen the main results of this paper .
\1 ) recall first two fundamental facts of tmt as a classical field theory : ( a ) the measure degrees of freedom appear in the equations of motion only via the scalar @xmath45 , eq.([zeta ] ) ; ( b ) the scalar @xmath45 is determined by the constraint which is nothing but a consistency condition of the equations of motion ( see eqs.([app1 ] ) , ( [ app3 ] ) and ( [ app4 ] ) in appendix b )
. therefore the constraint plays a key role in tmt .
note however that if we were ignore the gravity from the very beginning in the action ( [ totaction ] ) then instead of the constraint ( [ app4 ] ) we would obtain eq.([app1 ] ) ( where one has to put zero the scalar curvature ) .
in such a case we would deal with a different theory .
this notion shows that the gravity and matter intertwined in tmt in a much more complicated manner than in gr . hence introducing the new measure of integration @xmath12 we have to expect that the quantization of tmt may be a complicated enough problem .
nevertheless we would like here to point out that in the light of the recently proposed idea of ref.@xcite , the incorporation of four scalar fields @xmath461 together with the scalar density @xmath12 , eq.([phi ] ) , ( which in our case are the measure fields and the new measure of integration respectively ) , is a possible way to define local observables in the local quantum field theory approach to quantum gravity .
we regard this result as an indication that the effective gravity @xmath462 matter field theory has to contain the new measure of integration @xmath12 as it is in tmt .
\2 ) the assumption formulated in item 2 in sec.iia , that the measure fields @xmath461 ( or @xmath23 ) appear in the action ( [ s ] ) _ only _ via the measure of integration @xmath12 , has a key role in the tmt results and in particular for the resolution of the old cosmological constant problem , see sec.vi . in principle
one can think of breakdown of such a structure by quantum corrections .
however , fortunately there exists an infinite dimensional symmetry mentioned in item 2 of sec.iia which , as we hope , is able to protect the postulated structure of the action from a deformation caused by quantum corrections or at least to suppress such a quantum anomaly in significant degree
. therefore one can hope that the proposed resolution of the old cosmological constant problem holds in the quantized tmt as well .
\3 ) as we have noticed in sec.viia , the vacuum manifold ( [ min - v ] ) of the his is degenerate : the effective potential of the scalar sector remains equal to zero when shifting along this line in the @xmath442 plane .
however this degeneracy has no relation to any symmetry of the action , and in particular the kinetic energy of the scalar sector does not respect this shift as a symmetry .
therefore there are no reasons for quantum corrections not to break this shift symmetry of the vacuum manifold .
disappearance of the degeneracy of the vacuum manifold provides us with two important conclusions : a ) our model is out of danger to run against the problem of the appearance of a pseudo - goldstone boson ; b ) transition to the gauge symmetry broken phase will end in a certain point @xmath463 instead of a short interval of the line ( [ min - v ] ) as it was in the pure classical problem studied in sec.vii .
\4 ) recall finally that such quantum effects as the first and the second stages of particle creation discussed in details in sec.viie , supplement and strengthen the obtained classical results : the first stage of particle creation allows to hope that vev of the higgs field is independent of the initial conditions ; the second stage of particle creation should work as a mechanism of dissipation responsible for realization of the final state of the his in the gauge symmetry broken phase .
exploration of the quantization problems of tmt will be the subject of forthcoming research .
we thank l. amendola , s. ansoldi , r. barbieri , j. bekenstein , a. buchel , a. dolgov , s.b .
giddings , p. gondolo , j.b .
hartle , b - l .
. hung , d. kazanas , d. marolf , d.g .
mckeon , j. morris , v. miransky , h. nielsen , y.jack ng , h. nishino , e. nissimov , s. pacheva , l. parker , r. peccei , m. pietroni , s. rajpoot , r. rosenfeld , v. rubakov , e. spallucci , a. vilenkin , s. wetterich and a. zee for helpful conversations on different stages of this research .
our special gratitude to m. gedalin for his advice concerning some questions of dynamical systems and to l. prigozhin for great help in implementation of numerical solutions .
variation of the measure fields @xmath20 with the condition @xmath36 leads , as we have already seen in sec.ii , to the equation @xmath464 where @xmath27 is now defined , according to eq .
( [ s ] ) , as the part of the integrand of the action ( [ totaction ] ) coupled to the measure @xmath12 .
equation ( [ varphi ] ) in the context of the model ( [ totaction ] ) reads ( with the choice @xmath40 ) : @xmath465e^{\alpha\phi /m_{p } } -v_{1}e^{2\alpha\phi /m_{p } } + l_{fk } e^{\alpha\phi /m_{p } } \nonumber\\ & & -\frac{\upsilon}{\sqrt{2}}\left [ f_{n}\overline{n}n+f_{e}\overline{e}e\right ] = m^{4 } , \label{app1}\end{aligned}\ ] ] where the representation of the higgs field @xmath50 in the unitary gauge @xmath466 has been used in the last term of the l.h.s . of eq.([app1 ] ) ; to simplify the coming calculations it is convenient in the meanwhile not to do this explicitly with @xmath467 which includes interactions to gauge fields .
it can be noticed that the appearance of a nonzero integration constant @xmath468 spontaneously breaks the scale invariance ( [ stferm ] ) .
variation of the action ( [ totaction ] ) with respect to @xmath469 yields @xmath470 + ( \zeta + b_{\phi})e^{\beta}_{a}\phi_{,\mu}\phi_{,\beta } + \frac{1}{\sqrt{-g}}\frac{\partial(\sqrt{-g}l_{gauge } ) } { \partial e^{a,\mu } } e^{-\alpha\phi /m_{p } } \nonumber\\ & + & g_{\mu\beta}e^{\beta}_{a}\left[\frac{b_{g}}{\kappa}r(\omega , e ) -\frac{b_{\phi}}{2}g^{\mu\nu}\phi_{,\mu}\phi_{,\nu } -\frac{b_{g}}{2}g^{\mu\nu}(d_{\mu}h)^{\dag}d_{\nu}h + v_{2}e^{\alpha\phi /m_{p}}\right ] \nonumber\\ & + & g_{\mu\beta}e^{\beta}_{a}\left[-kl_{fk } + \frac{\upsilon}{\sqrt{2}}\left(h_{n}f_{n}\overline{n}n + h_{e}f_{e}\overline{e}e\right)e^{\frac{1}{2}\alpha\phi /m_{p } } \right ] \nonumber\\ & + & ( \zeta + b)\frac{i}{2 } \left[\overline{l}_{l } \left(\gamma^{a}\overrightarrow{d}_{\mu } -\overleftarrow{d}_{\mu}\gamma^{a}\right)l_{l } + \overline{e}_{r } \left(\gamma^{a}\overrightarrow{d}_{\mu } -\overleftarrow{d}_{\mu}\gamma^{a}\right)e_{r } + \overline{n}_{r } \left(\gamma^{a}\overrightarrow{d}_{\mu } -\overleftarrow{d}_{\mu}\gamma^{a}\right)n_{r } \right ] = 0
. \label{app2}\end{aligned}\ ] ] contraction of eq.([app2 ] ) with @xmath471 gives @xmath472 + ( \zeta -b_{\phi})\frac{1}{2}g^{\mu\nu}\phi_{,\mu}\phi_{,\nu } + 2v_{2}e^{\alpha\phi /m_{p } } + \frac{1}{2}(\zeta -3k)l_{fk } \nonumber\\ & + & \frac{2\upsilon}{\sqrt{2}}\left(h_{n}f_{n}\overline{n}n + h_{e}f_{e}\overline{e}e\right ) e^{\frac{1}{2}\alpha\phi /m_{p } } = 0 , \label{app3}\end{aligned}\ ] ] where the identity @xmath473 has been used .
excluding @xmath474 from eqs.([app1 ] ) and ( [ app3 ] ) we obtain the _ consistency condition _ of these two equations : @xmath475e^{\frac{1}{2}\alpha\phi /m_{p } } = 0 , \label{app4}\end{aligned}\ ] ] it is well known that the lagrangian for a single fermion field @xmath476 equals zero on the mass - shell .
this results from the linearity of the lagrangian both in @xmath476 and in @xmath477 .
the same is true for a system of fermion fields in our case because in spite of the presence of interactions , the fermion sector in ( [ totaction ] ) is linear in all fermion degrees of freedom .
therefore we have @xmath478e^{\frac{1}{2}\alpha\phi /m_{p } } \label{app5}\ ] ] inserting @xmath479 into the consistency condition , eq.([app4 ] ) , we get _ the constraint ( in the original frame ) _
@xmath480 where @xmath481 ( @xmath482 ) is the general notation for the primordial fermion fields @xmath53 and @xmath52 and @xmath483 , \qquad i = n , e .
\label{zetapm}\ ] ] contracting eq.([app2 ] ) with factor @xmath484 and using eqs.([app3 ] ) , ( [ app5 ] ) we get @xmath485 \nonumber\\ & + & \frac{\zeta + k}{\zeta + b_g } \frac{i}{2}e_{a\nu } \left[\overline{l}_{l } \left(\gamma^{a}\overrightarrow{d}_{\mu } -\overleftarrow{d}_{\mu}\gamma^{a}\right)l_{l } + \overline{e}_{r } \left(\gamma^{a}\overrightarrow{d}_{\mu } -\overleftarrow{d}_{\mu}\gamma^{a}\right)e_{r } + \overline{n}_{r } \left(\gamma^{a}\overrightarrow{d}_{\mu } -\overleftarrow{d}_{\mu}\gamma^{a}\right)n_{r } \right ] \nonumber\\ & + & g_{\mu\nu}\frac{\upsilon}{\sqrt{2}(\zeta + k)}e^{\frac{1}{2}\alpha\phi /m_{p } } \left[(h_{n}-k)f_{n}\overline{n}n+(h_{e}-k)f_{e}\overline{e}e\right ] .
\label{app12}\end{aligned}\ ] ] the scalar field @xmath0 equation of motion in the original frame can be written in the form @xmath486 \nonumber\\ & -&\frac{\alpha}{m_{p}}e^{\alpha\phi /m_{p}}\left[(\zeta
+ b_g)m^{4}e^{-\alpha\phi /m_{p}}+ [ ( b_g-\zeta)v_{1}-2v_{2}]e^{\alpha\phi /m_{p } } + ( b_g - b_{\phi})\frac{1}{2}g^{\alpha\beta}\phi_{,\alpha}\phi_{,\beta}\right ] \nonumber\\ & = & -\frac{\alpha}{m_{p}}e^{\frac{3}{2}\alpha\phi / m_{p } } \frac{\upsilon}{2\sqrt{2}(\zeta + k ) } \sum_{i = n , e}(\zeta-\zeta_1^{(i)})(\zeta-\zeta_1^{(2)})\overline{\psi}_{i}\psi_{i } , \label{phi - orig}\end{aligned}\ ] ] where eqs.([app1 ] ) and ( [ app5 ] ) have been used . in the vacuum of the gauge fields ,
the higgs field equation in the unitary gauge reads @xmath487 + e^{2\alpha\phi /m_{p}}[\zeta v^{\prime}_1+v^{\prime}_2]= -\frac{\upsilon}{\sqrt{2 } } \ , e^{\frac{3}{2}\alpha\phi / m_{p } } \sum_{i = n , e}f_i(\zeta + h_i)\overline{\psi}_{i}\psi_{i } , \label{higgs - orig}\ ] ] where @xmath488 , @xmath489 .
the gauge field @xmath49 equation in the unitary gauge and in the fermion vacuum reads @xmath490 + e^{\alpha\phi /m_{p } } ( \zeta + b_g)\sqrt{-g}g^{\mu\nu}\frac{1}{8}g^{\prime 2}\upsilon^2b_{\nu}=0 \label{gauge - orig}\ ] ] and similar for @xmath48 .
the fermion @xmath491 equation in the unitary gauge and for simplicity , in the gauge fields vacuum has the following form @xmath492\psi_i + \frac{ie^{-\alpha\phi /m_{p}}}{2(\zeta + k)\sqrt{-g}}\left[\partial_{\mu}\left((\zeta + k)\sqrt{-g}e^{\alpha\phi /m_{p}}e_a^{\mu}\right)\right]\gamma^a\psi_i \nonumber\\ & -&\frac{\zeta + h_i}{\zeta + k}e^{\frac{1}{2}\alpha\phi / m_{p}}\frac{f_i\upsilon}{\sqrt{2}}\psi_i=0 .
\label{ferm - orig}\end{aligned}\ ] ]
we present here what is the dependence of the spin connection @xmath58 on @xmath493 , @xmath45 , @xmath476 and @xmath477 .
varying the action ( [ totaction ] ) with respect to @xmath58 and making use that @xmath494 we obtain @xmath495 + \frac{\kappa}{4}\sqrt{-g}(\zeta + k)e^{c\mu}\varepsilon_{abcd}\overline{\psi } \gamma^{5}\gamma^{d}\psi=0 , \label{c2}\ ] ] where @xmath496 the solution of eq .
( [ c2 ] ) is represented in the form @xmath497 where @xmath498 is the riemannian part of the spin - connection , @xmath499 is the fermionic contribution that differs from the familiar one @xcite,@xcite by the factor @xmath500 and @xmath501 is the non - riemannian part of the spin - connection originated by specific features of tmt . in the einstein frame ,
i.e. in terms of variables defined by eq.([ctferm ] ) , the spin - connection read @xmath502 which is exactly the spin - connection of the einstein - cartan space - time@xcite with the vierbein @xmath503 .
in sec.iii we have seen that in the absence of fermions case ( and if @xmath158 ) the gravitational equations ( [ gef ] ) coincide with the einstein equations . analyzing eqs.([gef])-([muferm1 ] ) in more general cases it is easy to see that eqs.([gef ] ) and ( [ tmn ] ) are reduced to the einstein equations in the appropriate field theory model ( i.e. when the scalar field , electromagnetic field and massive fermions are sources of gravity ) if @xmath45 is constant and @xmath504 according to eqs.([tmn - noncan])-([muferm1 ] ) , in the case when a single massive fermion is a source of gravity , the condition ( [ noncan - ll - can ] ) is realized if @xmath505 where @xmath506 are defined in eqs.([zeta ] ) .
recall that existence of a noncanonical contribution to the energy - momentum tensor ( [ tmn - noncan ] ) , along with the @xmath45 dependence of the fermion mass ( [ muferm1 ] ) discovered in sec.iic , displays the fact that generically primordial fermion is very much different from the regular one ( see definitions in item ( ii ) of the introduction section ) . to answer the question what are the characteristic features of the regular massive fermion we have to take into account the undisputed fact that the classical tests of gr deal only with regular fermionic matter . hence
_ we should identify the regular fermions with states of the primordial fermions satisfying the condition _
( [ z-0 ] ) .
we are going now to understand the meaning of the constraint ( [ constraint2 ] ) .
we start from the detailed analysis of two limiting cases : ( a ) in space - time regions without fermions ; ( b ) in space - time regions occupied by regular fermions ; and afterwards we will be able to formulate the meaning of the constraint in general case .
we will proceed choosing the parameters @xmath507 , @xmath508 and/or @xmath218 as in sec.iv .
it is convenient to divide the analysis into a few steps : \(1 ) it follows from the condition @xmath509 that @xmath510 determined by the constraint in the absence of fermions case , eq.([zeta - without - ferm - delta=0 ] ) , has the same order of magnitude as the parameter @xmath218 .
\(2 ) recall that @xmath167 , eq.([veffvac - delta=0 ] ) , having the order of magnitude typical for the dark energy density ( in the absence of fermions case ) is obtained from @xmath511 , eq.([veff1 ] ) , as @xmath512 . therefore with the help of the item ( 1 )
we conclude that each time when @xmath45 has the order of magnitude close to that of the parameter @xmath218 ( and if no special tuning is assumed ) @xmath511 , eq.([veff1 ] ) , has the order of magnitude close to that of the dark energy density ( in the absence of fermions case ) .
\(3 ) it is easy to see that _ each time when _
@xmath45 _ has the order of magnitude close to that of the parameter
_ @xmath218 ( if no special tuning is assumed and in particular @xmath513 ) , the left hand side ( l.h.s . ) of the constraint ( [ constraint2 ] ) has the order of magnitude close to that of @xmath511 , eq.([veff1 ] ) , i.e. _ the l.h.s . of the constraint has the order of magnitude close to that of the dark energy density in the absence of fermions case_.
\(4 ) let us now turn to the right hand side ( r.h.s . ) of the constraint ( [ constraint2 ] ) in the presence of a single massive primordial fermion .
it contains factor @xmath514 , ( @xmath115 ) which have typical order of magnitude of the fermion canonical energy density @xmath515 .
if the primordial fermion is in a state of a regular fermion then according to the conclusion made at the end of the previous subsection , in the space - time region where the fermion is localized , the scalar @xmath45 must be @xmath516 ( or @xmath517 ) .
therefore in the space - time region occupied by a single regular fermion , the r.h.s . of ( [ constraint2 ] )
is @xmath518 due to our assumption that the dimensionless parameters @xmath218 , @xmath152 , @xmath2 , @xmath519 and @xmath520 have close orders of magnitude ( see paragraph after eq.([lgauge ] ) ) , it follows from the definitions ( [ zeta ] ) that both @xmath521 and @xmath522 have the order of magnitude close to that of @xmath218 .
hence in the space - time region occupied by a single regular fermion , the l.h.s . of ( [ constraint2 ] ) has the order of magnitude close to that of the dark energy density in the fermion vacuum .
it is evident that _ in normal particle physics conditions _
( see the last paragraph of sec.ii ) , that is when the energy density of a single fermion @xmath523 is tens of orders of magnitude larger than the fermion vacuum energy density , the balance dictated by the constraint is satisfied in the present day universe just due to the condition ( [ z-0 ] ) .
5 in more general cases , i.e. when primordial fermion is in a state different from the regular one , the meaning of the constraint is similar : _ the balance between the scalar dark energy contribution to the l.h.s . of the constraint and the fermionic contribution to the r.h.s . of the constraint _
is realized due to the factors @xmath113 . in other words , _ the constraint describes the local balance between the fermion energy density and the scalar dark energy density _ in the space - time region where the wave function of the primordial fermion is not equal to zero ; _ by means of this balance the constraint determines the scalar _ @xmath524 .
note also that due to this balance , _ the degree of localization of the fermion and values of @xmath524 may be strongly interconnected_. one can suggest the following two alternative approaches to the question of how a primordial fermion can be realized as a regular one : _ the first approach _ discussed in refs.@xcite , @xcite , is based on the idea of the `` maximal economy '' .
we start from one primordial fermion field for each type of fermions : one neutral primordial lepton field @xmath53 , one charged primordial lepton field @xmath52 and similar for quarks . in other words
we start from one generation of fermions .
splitting of the primordial fermions into families occurs only in normal particle physics conditions , i.e. when the fermion energy density is huge in comparison with the vacuum energy density .
one of the possibilities for this to be realized is the above mentioned condition @xmath525 .
the appropriate two constant solutions for @xmath45 , i.e. @xmath526 , correspond to two different states of the primordial fermions with _ different constant masses _
determined eqs.([muferm1 ] ) where we have to substitute @xmath506 instead of @xmath45 .
so , in the normal particle physics conditions , the scalar @xmath45 plays the role of an additional degree of freedom determining different mass eigenstates of the primordial fermions which we want to identify with different fermion generations .
note that the classical tests of gr deal in fact with matter built of the fermions of the first generation ( may be with a small touch of the second generation ) .
this is why one can identify the states of the primordial fermions realized as @xmath527 with the first two generations of the regular fermions .
for example , if the free primordial electron @xmath52 is in the state with @xmath528 ( or @xmath529 ) , it is detected as the regular electron @xmath530 ( or muon @xmath531 ) and similar the primordial neutrino @xmath53 splits into the regular electron and muon neutrinos with masses respectively : @xmath532 it turns out that there is only one more additional possibility to satisfy the constraint ( [ constraint2 ] ) when primordial fermion is in the normal particle physics conditions .
this is the solution @xmath533 which one can associate with the third generation of fermions ( for details see refs.@xcite , @xcite ) .
it is interesting that in contrast to the first two generations , the third generation defined by this way , may have gravitational interaction with unusual features since the condition ( [ noncan - ll - can ] ) may not hold .
the described splitting of the primordial fermions into three generations in the normal particle physics conditions is the family replication mechanism proposed in refs.@xcite , @xcite . _
the second approach _ is based on the idea that the three families of fermions of the standard model exist from the beginning in the original action , i.e. not to use the family replication mechanism for explanation of the observed three generations of fermions . in this case
again , exactly as it was in the first approach , the primordial fermions turn into the regular fermions only in the normal particle physics conditions .
now however if we interpret the state of the primordial fermions with , for example , @xmath534 as the observable regular fermions , then some role should be assigned to the states with @xmath535 and @xmath536 . by means of a choice of the parameters one can try for example to provide very large masses of the regular fermions with @xmath535 and @xmath536 that might explain the reason why they are unobservable so far .
however these questions are beyond of the goals of this paper and together with many other aspects of fermions in tmt will be studied in a separate publication .
reproducing einstein equations when the primordial fermions are in the states of the regular fermions is not enough in order to assert that gr is reproduced .
the reason is that at the late universe , as @xmath537 , the scalar field @xmath0 effective potential is very flat and therefore due to the yukawa - type coupling of massive fermions to @xmath0 , ( the r.h.s . of eq.([phief+ferm1 ] ) ) , the long range scalar force appears to be possible in general .
the yukawa coupling `` constant '' is @xmath538 .
applying our analysis of the meaning of the constraint in two previous subsections , it is easy to see that for regular fermions with @xmath539 the factor @xmath113 is of the order of the ratio of the vacuum energy density to the regular fermion energy density .
thus we conclude that the 5-th force is extremely suppressed for the fermionic matter observable in classical tests of gr .
it is very important that this result is obtained automatically , without tuning of the parameters and it takes place for both approaches to realization of the regular fermions in tmt discussed in previous subsection .
the results of the previous three sections depend very much on the choice of the parameters @xmath87 , @xmath88 and @xmath242 in the action ( [ totaction ] ) .
let us recall that the curvature term in the action ( [ totaction ] ) couples to the measure @xmath540 while the @xmath0 kinetic term couples to the measure @xmath541 .
this is the reason of @xmath160 .
if we were choose the fine tuned condition @xmath542 then both the curvature term and the @xmath0 kinetic term would be coupled to the same measure @xmath543 .
one can also pay attention that depending on the choice of one of the alternative conditions @xmath171 or @xmath171 we realize different shapes of the effective potential if @xmath170 ( see fig.1 ) . and again , if instead we were choose the fine tuned condition @xmath185 then the action would contain only one prepotential coupled to the measure @xmath540 .
so , in order to avoid fine tunings we have introduced asymmetries in the couplings of the different terms in the lagrangian densities @xmath27 and @xmath16 to measures @xmath12 and @xmath13 . in order to display the role of these asymmetries
it is useful to consider what happens if such asymmetries are absent in the action at all . in other words
we want to study the gravity+dilaton model where both @xmath158 and @xmath185 .
in such a case the action contains only one lagrangian density coupled to the measure @xmath543 : @xmath544 where @xmath545 .
an equivalent statement is that @xmath546 ; it is an example of the very special class of the tmt models where @xmath27 is proportional to @xmath16 . to see the cosmological dynamics in this model one can use the results of sec.iiib .
if we assume in addition @xmath169 and @xmath184 , then after the shift @xmath547 where @xmath548 ( which is not a shift symmetry in this case ) , the effective potential ( [ veffvac - delta=0 ] ) takes the form @xmath549 in contrast to general cases ( @xmath550 ) this potential has no flat regions and it is symmetric around a certain point in the @xmath0-axis .
this form of the potential ( with an additional constant ) has been used in a model of the early inflation@xcite .
although we have assumed in the main text that @xmath27 and @xmath16 are @xmath20 independent , a contribution equivalent to the term @xmath551 can be effectively reproduced in the action ( [ s ] ) if a nondynamical field ( lagrange multipliers ) is allowed in the action . for this purpose
let us consider the contribution to the action of the form @xmath552d^{4}x \label{l - m - action}\ ] ] where @xmath296 is an auxiliary nondynamical field and @xmath553 is an analytic function .
varying @xmath296 we obtain @xmath554 that can be solved for @xmath296 : @xmath555 where @xmath556 is the inverse function of @xmath557 . inserting this solution for @xmath296 back into the action ( [ l - m - action ] ) we obtain @xmath558 where the auxiliary field has disappeared and @xmath559 to see the difference between effect of this type of auxiliary fields as compared with a model where the @xmath296 field is equipped with a kinetic term ,
let us consider two toy models including gravity and @xmath296 field : one - without kinetic term @xmath560d^{4}x \label{toy-1}\ ] ] and the other - with a kinetic term @xmath561d^{4}x \label{toy-2}\ ] ] where @xmath562 is a real constant .
for both of them it is assumed the use of the first order formalism .
the first model is invariant under local transformations @xmath563 , @xmath564 , @xmath565 where @xmath566 is an arbitrary space - time function while in the second model the same symmetry transformations hold only if @xmath566 is constant .
variation of the measure fields @xmath20 in the model ( [ toy-2 ] ) leads ( if @xmath36 ) to @xmath567 where @xmath468 is the integration constant . on the other hand varying the action ( [ toy-2 ] ) with respect to @xmath568 gives @xmath569 where @xmath570 . the corresponding equation in the model ( [ toy-1 ] )
is obtained from ( [ toy-1-gmunu ] ) by omitting the term with gradients of @xmath296 .
it follows from eqs.([toy-1-phi ] ) and ( [ toy-1-gmunu ] ) that @xmath571 this result holds in both model . in the model ( [ toy-1 ] ) , variation of @xmath296 results in @xmath572 which is consistent with eq.([toy - constr ] ) only if the integration constant @xmath573 .
this means that through the classical mechanism displayed in tmt , _ it is impossible to achieve spontaneous breakdown of the local scale invariance _
the first model possesses .
transition to the einstein frame where the space - time becomes riemannian is implemented by means of the conformal transformation @xmath574 . for the model ( [ toy-1 ] ) the gravitational equations in the einstein frame read @xmath575
this means that the model ( [ toy-1 ] ) with auxiliary ( nondynamical ) field @xmath296 intrinsically contains a constant vacuum energy . in the model ( [ toy-2 ] ) , where @xmath296 appears as a dynamical field , the gravitational equations in the einstein frame results from eq.([toy-1-gmunu ] ) @xmath576
it is convenient to rewrite this equation in terms of the scalar field @xmath577 : @xmath578 the @xmath0-equation reads @xmath579 .
similar to the general discussion in the main text we see that if the @xmath296 field is dynamical then tmt provides the vacuum with zero energy without fine tuning .
however it is very unlikely that a nondynamical scalar field will not acquire a kinetic term after quantum corrections@xcite .
then it becomes dynamical which restores the above results for the model ( [ toy-2 ] ) .
this is why we have ignored the rather formal possibility of introducing the nondynamical scalars into the fundamental action of the models studied in this paper .
a. g. riess et al .
, _ astron .
j. _ * 116 * , 1009 ( 1998 ) ; s. perlmutter _
_ , _ astrophys
. j. _ * 517 * , 565 ( 1999 ) ; n. bahcall , j.p .
ostriker , s.j . perlmutter and p.j .
steinhardt , _ science _ * 284 * , 1481 ( 1999 ) ; d.n .
spergel , _ et al . _
[ wmap collaboration ] , _ astrophys .
148 _ , * 175 * ( 2003 ) ; a. c. s. readhead , _ et al . _
. j. _ * 609 * , 498 ( 2004 ) ; j. h. goldstein et al . , _
_ 599 , 773 - 785 ( 2003 ) ; r. rebolo , _ et al .
_ , astro - ph/0402466 ; m. tegmark , _ et al .
_ _ phys.rev . _ * d69 * , 103501 ( 2004 ) ; e. hawkins _ et al . , _
_ mon.not.roy.astron.soc . _ * 346 * , 78 ( 2003 ) ; w. l. freedman , _ et al .
, astrophys.j .
_ * 553 * , 47 ( 2001 ) ; r. daly , s.g .
djorgovski , _ astrophys.j . _ * 597 * , 9 , ( 2003 ) . c. wetterich , nucl
* b302 * , 668 ( 1988 ) . b. ratra and p.j.e .
peebles , phys . rev . *
d37 * , 3406 ( 1988 ) ; p.j.e .
peebles and b. ratra , astrophys .
j. * 325 * , l17 ( 1988 ) .
r. caldwell , r. dave and p. steinhardt , phys .
lett . * 80 * , 1582 ( 1998 ) ; n. weiss , phys . lett .
* b197 * , 42 ( 1987 ) ; y. fujii and t. nishioka , phys . rev . *
d42 * , 361 ( 1990 ) ; m.s .
turner and m. white , phys . rev .
* d56 * , r4439 ( 1997 ) ; p. ferreira and m.joyce , phys .
lett . * 79 * , 4740 ( 1997 ) ; phys . rev . *
d58 * , 023503 ( 1998 ) .
e. copeland , a. liddle and d. wands , phys .
d57 , 4686 ( 1998 ) ; p. steinhardt , l wang and i. zlatev , phys . rev . *
d59 * , 123504 ( 1999 ) .
t. chiba , t.okabe and m yamaguchi , phys.rev.d*62 * 023511 ( 2000 ) ; c. armendariz - picon , v. mukhanov and p.j .
steinhardt , phys.rev.lett . * 85 * 4438 , 2000 ; phys.rev.d*63 * 103510 ( 2001 ) ; j.k .
erickson , r.r .
caldwell , p.j .
steinhardt , c. armendariz - picon , v. mukhanov , phys.rev.lett . * 88 * 121301 ( 2002 ) ; t. chiba , phys.rev.d*66 * 063514 ( 2002 ) .
j.a.casas , j. garcia - bellido and m. quiros , _ class .
* 9 * , 1371 ( 1992 ) ; g.w .
anderson and s.m .
carroll , astro - ph/9711288 ; d. comelli , m. pietroni and a. riotto , _ phys .
lett . _ * b571 * , 115 ( 2003 ) .
chimento , a.s .
jakubi and d. pavon , phys .
rev . d*62 * , 063508 ( 2000 ) ; w. zimdahl , d. schwarz , a. balakin and d. pavon phys .
rev . d*64 * , 063501 ( 2001 ) ; l.p .
chimento , a.s .
jakubi , d. pavon and w. zimdahl , phys .
d*67*,083513 ( 2003 ) ; a.a .
sen and s. sen , mod .
a * 16 * , 1303 ( 2001 ) .
holden and d. wands , _ phys .
* d61 * , 043506 ( 2000 ) ; a.p .
billyard and a.a .
coley , _ phys .
* d61 * , 083503 ( 2000 ) ; n. bartolo and m. pietroni , _ phys .
rev . _ * d61 * , 023518 ( 2000 ) ; m. gasperini , _ phys .
rev . _ * d64 * , 043510 ( 2001 ) ; a. albrecht , c.p .
burges , f. ravndal and c. skordis , astro - ph/0107573 ; l.p .
chimento , a.s . jakubi and d. pavon , _ phys .
_ * d62 * , 063508 ( 2000 ) ; w. zimdahl , d. schwarz , a. balakin and d. pavon _ phys .
_ * d64 * , 063501 ( 2001 ) ; l.p .
chimento , a.s .
jakubi , d. pavon and w. zimdahl , _ phys .
_ * d67*,083513 ( 2003 ) ; a.a . sen and s.
sen , _ mod .
* a16 * , 1303 ( 2001 ) ; g.r .
farrar and p.j.e .
peebles , astro - ph/0307316 m ; axenides , k. dimopoulos , _ jcap _ * 0407 * , 010,2004 ; m. nishiyama , m. morita , m. morikawa , astro - ph/0403571 ; r. catena , m. pietroni and l. scarabello , _ phys.rev . _
* d70 * , 103526 ( 2004 ) .
kamenshchik , u. moschella and v. pasquier , phys .
b*511 * , 265 ( 2001 ) ; n. bilic , g.b .
tupper and r.d .
viollier , phys .
b*535 * , 17 ( 2002 ) ; m.c .
bento , o. bertolami and a.a .
sen , phys .
d*66 * ( 2002 ) ; r.r .
caldwell , phys.lett . b*545 * , 23 ( 2002 ) ; g.w .
gibbons , hep - th/0302199 ; p. singh , m. sami , n. dadhich , phys.rev . d*68 * 023522 ( 2003 ) ; m. sami , a. toporensky , mod.phys.lett .
a*19 * , 1509 ( 2004 ) ; p.f .
gonzalez - diaz , phys.lett .
b*586 * 1 ( 2004 ) ; a. vikman , phys.rev .
d*71 * , 023515 ( 2005 ) ; p.f .
gonzalez - diaz , c.l .
siguenza nucl.phys .
b*697 * 363 2004 ; wayne hu , phys.rev .
d*71 * 047301 ( 2005 ) ; r .- g .
cai , a. wang , jcap 0503 ( 2005 ) ; s. nojiri , s. d. odintsov , s. tsujikawa , phys.rev . d*71 * 063004 ( 2005 ) ; e. elizalde , s. nojiri , s.d .
odintsov , p. wang , phys.rev . d*71 * 103504 ( 2005 ) ; h. stefancic , phys.rev . d*71 * 124036 ( 2005 ) ; l. amendola , s. tsujikawa , m. sami , phys.lett . b*632 * , 155 ( 2006 ) .
gibbons , phys .
b*537 * , 1 ( 2002 ) ; m. fairbairn and m.h.g .
tytgat , phys.lett.b*546 * , 1 ( 2002 ) ; s. mukohyama , phys . rev .
d*66 * , 024009 ( 2002 ) ; a.v .
frolov , l. kofman , a.a .
starobinsky , phys.lett . b*545 * , 8 ( 2002 ) ; y .- s .
piao , r .-
cai , x. zhang and t .- z .
zhang , phys .
rev . d*66 * , 121301 ( 2002 ) ; t. padmanabhan and t.r .
chounhury , phys .
rev . d*66 * , 081301 ( 2002 ) ; a. feinstein , phys.rev .
d*66 * , 063511 ( 2002 ) .
l. randall and r. sundrum , phys .
lett . * 83 * , 4690 ( 1999 ) ; m. gogberashvili , int.j.mod.phys .
d*11 * , 635 ( 2002 ) ; t. shiromizu , k. maeda and m. sasaki , phys .
rev . d*62 * , 024012 ( 2000 ) ; r. maartens , d. wands , b. bassett and i. heard , phys .
rev . d*62 * 041301 ( 2000 ) ; h. collins and b. holdom , phys .
d*62 * , 105009 ( 2000 ) ; c. deffayet , g. dvali and g. gabadadze , phys .
d*65 * , 044023 ( 2002 ) ; v. sahni and y. shtanov , jcap * 0311 * , 013 ( 2003 ) .
s. capozziello , v.f .
cardone , s. carloni and a. troisi , int.j.mod.phys .
* d12 * 1969 ( 2003 ) ; a. d. dolgov and m. kawasaki , phys .
lett * b573 * , 1 ( 2003 ) ; s.m.carroll , a. de felice , v. duvvuri , d.a .
easson , m. trodden and m.s .
turner , phys.rev .
* d71 * 063513 ( 2005 ) .
guendelman and a.b .
kaganovich , _ phys .
rev . _ * d53 * , 7020 ( 1996 ) ; _ mod .
_ * a12 * , 2421 ( 1997 ) ; _ phys .
rev . _ * d55 * , 5970 ( 1997 ) ; _ mod .
* a12 * , 2421 ( 1997 ) ; _ phys .
rev . _ * d55 * , 5970 ( 1997 ) ; ibid .
* d57 * , 7200 ( 1998 ) ; _ mod .
_ * a13 * , 1583 ( 1998 ) .
guendelman , _ mod .
lett . _ * a14 * , 1043 ( 1999 ) ; _ class .
* 17 * , 361 ( 2000 ) ; gr - qc/9906025 ; _ mod .
_ * a14 * , 1397 ( 1999 ) ; gr - qc/9901067 ; hep - th/0106085 ; _ found .
_ * 31 * , 1019 ( 2001 ) ; e.i .
guendelman , class.quant.grav .
* 17 * , 3673 ( 2000 ) ; e.i .
guendelman , phys .
rev . d*63 * , 046006 ( 2002 ) ; e.i .
guendelman , a.b .
kaganovich , e. nissimov , s. pacheva , phys .
rev . d*66 * , 046003 ( 2002 ) ; hep - th/0210062 ; hep - th/0304269 ; e.i .
guendelman , a. kaganovich , e. nissimov , s. pacheva , phys.rev . d*72 * , 086011 ( 2005 ) .
abbott , m.b .
wise , nucl.phys .
b*244 * , 541 ( 1984 ) ; f. lucchin , s. matarrese , phys.rev . d*32 * , 1316 ( 1985 ) ; j.d .
barrow , phys.lett . b*187 * , 12 ( 1987 ) ; j. yokoyama , kei - ichi maeda , phys.lett .
b*207 * , 31 ( 1988 ) ; a.r .
liddle , phys.lett .
b*220 * , 502 ( 1989 ) ; v. muller , h.j .
schmidt , a.a .
starobinsky , class.quant.grav . * 7 * , 1163 ( 1990 ) ; b. ratra , phys.rev . d*45 * , 1913 ( 1992 ) .
bennett , h.b .
nielsen , i. picek , phys.lett . b*208 * , 275 ( 1988 ) ; d.l .
bennett , h.b .
nielsen , int.j.mod.phys . a*9 * , 5155 ( 1994 ) ; c.d .
froggatt , h.b .
nielsen , phys.lett . b*368 * , 96 ( 1996 ) ; _ origin of symmetries _ , world scientific , singapore ( 1991 ) . | we study the scalar sector of the two measures field theory ( tmt ) model in the context of cosmological dynamics .
the scalar sector includes the dilaton @xmath0 and the higgs @xmath1 fields .
the model possesses gauge and scale invariance .
the latter is spontaneously broken due to intrinsic features of the tmt dynamics .
the scalar sector dynamics represents an explicit example of @xmath2-essence resulting from first principles where @xmath0 plays the role of the inflaton . in the model with the inflaton @xmath0 alone , in different regions of the parameter space
the following different effects can take place _ without fine tuning of the parameters and initial conditions _ :
a ) possibility of a power law inflation driven by the scalar field @xmath0 which is followed by the late time evolution driven both by a small cosmological constant and the scalar field @xmath0 with a quintessence - like potential ; smallness of the cosmological constant can be achieved without fine tuning of dimensionfull parameters .
b ) possibility of _ resolution of the old cosmological constant problem _ : this is done in a consistent way hinted by s. weinberg in his comment concerning the question of _ how one can avoid his no - go theorem_. c ) the power law inflation without any fine tuning may end with damped oscillations of @xmath0 around the state with zero cosmological constant .
d ) there are regions of the parameters where the equation - of - state @xmath3 in the late time universe is @xmath4 and @xmath5 _ asymptotically _ ( as @xmath6 ) _ approaches _ @xmath7 _ from below_. this effect is achieved without any exotic term in the action . in a model with both @xmath0 and @xmath1 fields , a scenario which resembles the hybrid inflation is realized but there are essential differences , for example : the higgs field undergos transition to a gauge symmetry broken phase @xmath8 soon _ after _ the end of a power law inflation ; there are two oscillatory regimes of @xmath1 , one around @xmath9 at 50 e - folding before the end of inflation , another - during transition to a gauge symmetry broken phase where the scalar dark energy density approaches zero without fine tuning ; the gauge symmetry breakdown is achieved _ without tachyonic mass term _ in the action . |
quantum entanglement plays a central role in the emerging fields of quantum computation@xcite , quantum cryptography@xcite , quantum teleportation@xcite , dense coding@xcite and quantum communication@xcite .
the characterisation of entanglement is a challenging problem@xcite and considerable effort has been invested in characterising entanglement in a variety of contexts@xcite .
one such context , quantum teleportation , has played a crucial role in understanding how entanglement can be used as a resource for communication .
the recent experimental demonstrations@xcite suggest that quantum teleportation could be viewed as an achievable experimental technique to quantitatively investigate quantum entanglement .
teleportation is a way of transmitting an unknown quantum state to a distant receiver with far better reliability than can be achieved classically . as the entanglement of the enabling resource
is degraded , the fidelity of the teleportation protocol is diminished .
in this paper we attempt to make this intuition more precise using specific examples from quantum optics .
three entangled resources are considered : states with fixed total photon number , number states entangled at a beam splitter , and the two - mode squeezed vacuum state@xcite .
the examples we discuss exhibit quantum correlations between the photon number in each mode and , simultaneously , between the phase of each mode . in reality
, the entanglement will not be perfect , but degraded to some extent by uncontrolled interactions with an environment during formation .
to model this we consider phase fluctuations on each mode independently . in the limit of completely random phase
we are left with only the classical intensity ( photon number ) correlations .
the state is no longer entangled and the fidelity of the protocol depends only on the classical intensity correlations remaining in the resource .
intuitively entanglement refers to correlations between distinct subsystems that can not be achieved in a classical statistical model . of course correlations
can exist in classical mechanics , but entanglement refers to a distinctly different kind of correlation at the level of quantum probability amplitudes .
the essential difference between quantum and classical correlations can be described in terms of the separability of states@xcite .
a density operator of two subsystems is separable if it can be written as the convex sum@xcite @xmath0 where @xmath1 and @xmath2 are density matrices for the two subsystems and the @xmath3 are positive weights satisfying @xmath4 .
for example , for two harmonic oscillators ( @xmath5 and @xmath6 ) the density operator which has correlated energy @xmath7 is separable , ( where we use the notation @xmath8 ) while the pure state @xmath9 has the same classical correlation but is not separable . in this form
we see that is possible for a separable and an entangled state to share similar classical correlations for some variables .
consider a communication protocol in which the results of measurements made on a physical system are transmitted to a distant receiver .
the goal of the receiver is to reconstruct the physical state of the source , using only local resources , conditioned on the received information .
the communication that takes place is of course entirely classical . in a teleportation protocol
there is one additional feature : quantum correlations , entanglement , are first shared between the sending and receiving station .
the degree of entanglement shared by sender and receiver is called the teleportation _
resource_. if there is no shared quantum correlation between the sender and receiver , the protocol is called classical .
the extent and nature of the quantum correlations in the resource determine the fidelity of the protocol . under ideal conditions
the unknown state of some physical system at the transmitting end can be perfectly recreated in another physical system at the receiving end .
there are many ways in which actual performance can differ from the ideal . in this paper
we analyse the change in the performance of teleportation protocols as the degree of entanglement in the resource is varied by decoherence .
our primary objective is to use the fidelity of a teleportation protocol to compare and contrast different kinds of entangled oscillator states . a general teleportation protocol proceeds as follows .
the sender , alice , has a _
state , @xmath10 , she wishes to teleport to bob , the receiver .
alice and bob each have access to one part of an entangled bipartite physical system prepared in the state @xmath11 . in this paper
the bipartite physical system is a two - mode electromagnetic field . in order to send the state of the target to bob
, alice performs a joint measurement on the target and her mode .
she then sends the information gained from these measurements to bob via a classical channel .
bob performs local unitary transformations on the mode in his possession according to the information alice sends to him , thereby attempting to recreate the initial target state .
we quantify the quality of the protocol by the probability that bob s received state is the same as the target state .
this quantity is known as the _
fidelity_. the fidelity of quantum teleportation protocol is determined by the degree of shared entanglement , the quality of the measurements made by the sender , the quality of the classical communication channels used and how well bob can implement the desired unitary transformations . in this paper
we will discuss only the first of these ; the amount of shared entanglement . in the original teleportation protocol@xcite ,
the bipartite system was made up of two systems each described by a two dimensional hilbert space , that is to say , two qubits , and the shared entangled state was a _ maximally _
entangled state@xcite . in the case of two correlated harmonic oscillators , or two field modes
, we can not define maximal entanglement in quite the same way , as the entropy of each component system can be arbitrarily large . in this paper
we define extremal entangled pure states of two field modes in terms of the total mean photon number and the total maximum photon number . in the case of a system with a finite dimensional hilbert space , a state of maximum ( von neumann )
entropy is simply the identity operator in that hilbert space .
a natural generalisation of this idea to infinite hilbert spaces would define a maximum entropy state subject to some constraint , such as mean energy or total energy .
these of course define the canonical ensemble and micro canonical ensemble of statistical mechanics . in the case of
entangled pure states the araki - lieb@xcite inequality indicates that the entropy of each component system is equal .
as the entropy of a harmonic oscillator scales with mean energy , this indicates that each component subsystem has the same mean energy .
if we maximise the entropy of each subsystem subject to a constraint on the mean energy , the state must be a thermal state .
the entangled pure two mode state for which the reduced density operator of each mode is thermal , is the squeezed vacuum state .
@xmath12 the mean photon number in each mode is given by @xmath13 .
if , however , we constrain the total photon number , @xmath14 , of each mode we get a very different expression for a maximally entangled state , @xmath15 the entropy of the reduced state of each mode is @xmath16 while the mean photon number is @xmath17 .
while squeezed vacuum states may be achieved in the laboratory , states with fixed total photon number have not been produced , and will not be possible until we have a reliable @xmath14 photon source .
there are now a couple of proposals for such sources@xcite and it may not be too long before they are used in teleportation schemes . teleportation fidelity for infinite dimensional hilbert spaces must necessarily vary from unity for an arbitrary target state , as the notion of a maximally entangled resource differs from the finite dimensional case
. the teleportation protocol can also be degraded by unknown incoherent processes that corrupt the purity of the shared entanglement .
of course in some cases these incoherent processes may destroy the correlations entirely , for example by absorbing all the photons in each mode before alice and bob get to use them . in this paper , however , we will only consider those decoherence processes that change the purity of the states and leave unchanged the classical intensity correlations .
in a recent paper by milburn and braunstein@xcite a teleportation protocol was presented using joint measurements of the photon number difference and phase sum on two field modes .
this protocol is possible because the number difference and phase sum operators commute , thus allowing determination of these quantities simultaneously and to arbitrary accuracy .
number _ sum _ and phase _ difference _ operators also commute , implying that if eigenstates of these operators can be found then a teleportation protocol is possible .
such a protocol is discussed below .
recently , teleportation using number sum and phase difference measurements was described @xcite .
that work did not address how the degree of entanglement in the resource changes the teleportation fidelity as we do here . because the number sum and phase difference operators commute we look for simultaneous eigenstates of these observables .
consider states of the form @xmath18 which are eigenstates of number sum with eigenvalue @xmath14 .
the labels @xmath19 and @xmath20 refer to the sender s and receiver s component of the entangled modes respectively , and the @xmath21 satisfy @xmath22 . this state will be maximally entangled when the @xmath23 are all equal , giving the resource , @xmath24 this state tends towards eigenstates of phase difference as @xmath25 . to see this
consider the joint phase probability density of eq .
( [ eq : idealres ] ) , which is determined by the ideal joint phase operator projection operator @xmath26 as @xmath27 substituting the state in eq .
( [ eq : genideal ] ) we have @xmath28 where @xmath29 . the probability density as a function of @xmath14 and @xmath30
is shown in fig .
[ fig : phasedist ] , and indicates that the density becomes sharply peaked about @xmath31 in the interval @xmath32 $ ] as @xmath14 gets larger . hence the states of eq .
( [ eq : idealres ] ) tend to eigenstates of phase difference with increasing @xmath14 .
the state to be teleported , the target state , can be written in the general form @xmath33 the input state to the protocol is then @xmath34 if alice measures the number sum of the target and her component of the entangled resource ( i.e. @xmath35 ) with result @xmath36 , the conditional state of the total system is @xmath37^{-1/2}\sum_{n } c_{q - n+n } { | q - n+n \rangle_{t}}{| n - n \rangle_{a}}{| n \rangle_{b } } , \label{eq : condqideal}\ ] ] where @xmath38 runs from @xmath39 to @xmath14 . the probability of obtaining the result @xmath36 is @xmath40 the subscript @xmath41 emphasises that this probability refers to the idealised resource .
alice now measures the phase difference with result @xmath30 .
the conditional state of bob s mode is then the pure state @xmath42^{-1/2}\sum_{n } e^{2in\phi_- } c_{q - n+n }
{ | n \rangle_{b}}. \label{eq : bobideal}\ ] ] using the results @xmath36 and @xmath30 , and knowledge of the number of fock states in the resource ( @xmath14 ) , bob has sufficient information to reproduce the target state .
he does this by amplifying his mode so that @xmath43 and phase shifting it by @xmath44 .
the unitary amplification operation is described in @xcite .
these operations complete the protocol and the state bob finally has in his possession is @xmath45^{-1/2 } \sum_{n } c_{q - n+n } { | q - n+n \rangle_{b}}. \label{eq : outideal}\ ] ] the fidelity of this protocol depends on the result @xmath36 and is @xmath46 as the fidelity depends on the result of the number sum measurement it varies from one run to the next . to obtain an overall figure of merit for the protocol we define the average fidelity , @xmath47 in this case we find @xmath48 to see
how well the teleportation protocol performs , we shall consider some examples .
let the target state be a number state , @xmath49 so the only coefficient available is @xmath50 , which is one .
we find that the teleportation fidelity is unity , independent of the measurement of @xmath36 , because the only term appearing in the summations of both the fidelity and the probability is that corresponding to @xmath50 .
hence , this protocol works perfectly if the target is a number state . in fig .
[ fig : fideal ] we show the average fidelity as a function of the total photon number in the resource for a coherent state target with amplitude @xmath51 .
it is clear that increasing the number of photons in the entangled resource improves the teleportation protocol .
the resource states discussed in sec . [ sec : ideal ] illustrate the protocol well , but are not produced by any known physical interaction .
however , the beam splitter interaction can be shown to give a resource with similar properties to eq .
( [ eq : idealres ] ) .
the beam splitter interaction is described by @xcite @xmath52 where the operators @xmath5 , @xmath53 , @xmath6 and @xmath54 are the usual boson annihilation and creation operators for modes @xmath19 and @xmath20 .
@xmath14 being the number of photons at each input port of the beam splitter . because the number sum of the two modes is a constant ( @xmath55 )
we can rewrite the resource in terms of eigenstates of number sum .
the resource is now written as @xmath56 the coefficients @xmath57 are derived by first using schwinger s boson representation of angular momentum , with total angular momentum quantum number @xmath58 , and then identifying these coefficients as rotation matrix elements@xcite .
the @xmath57 coefficients are defined by @xmath59 where @xmath60^{\frac{1}{2}}}\nonumber\\ & & \times \sum_s \frac{(-1)^{m'-m+s } \left(\cos\frac{\beta}{2}\right)^{2j+m - m'-2s } \left(\sin\frac{\beta}{2}\right)^{m'-m+2s}}{(j+m - s)!s!(m'-m+s)!(j - m'-s)!}. \nonumber\\ & & \label{eq : ddefnexplicit}\end{aligned}\ ] ] the variable @xmath61 ranges over all integer values where the factorials are non - negative@xcite .
it is easy to verify that all coefficients with @xmath38 odd are zero .
this protocol proceeds identically to that discussed in sec .
[ sec : ideal ] .
we illustrate this variation of the protocol with the pure state form of the resource as given in eq .
( [ eq : bsresfinal ] ) . after a number sum and phase difference measurement on modes @xmath62 and @xmath19 , and then applying the amplification @xmath63 and phase shift @xmath64 , the output state becomes , @xmath65^{-1/2 } \sum_{n = 0}^{\mathrm{min}(q,2n ) } c_{q - n } d_{n - n } { | q - n \rangle_{b } } , \label{eq : bsout}\ ] ] where the probability for a number sum result @xmath36 is @xmath66 the teleportation fidelity
is found to be @xmath67 if we again consider a coherent state target of amplitude @xmath51 , we can compare the beam splitter generated resource with the ideal resource in sec .
[ sec : ideal ] .
the average fidelity as a function of energy in the resource is shown in fig .
[ fig : fbscohstate ] and is almost identical to fig .
[ fig : fideal ] except that its maximum is approximately one half as opposed to unity .
this is due to the fact that all terms in eq .
( [ eq : bsresfinal ] ) with @xmath38 odd are zero .
effectively only half of the perfect correlations in the ideal entangled resource are available , hence the maximum fidelity we would expect under such circumstances is @xmath68 .
even so , the state becomes a better teleportation resource with increasing @xmath14 ( see fig .
[ fig : fbscohstate ] ) .
it is , however , possible to teleport those states which have no odd photon number components with near unit fidelity .
an example is the even `` cat '' state , formed from the superposition of two coherent states of equal real amplitude but opposite sign @xcite , @xmath69 the average fidelity in this case is shown in fig .
[ fig : fbscatstate ] .
this result implies that it may be possible to tailor resources for given applications so that certain classes of states may be teleported well , without necessarily being able to teleport an arbitrary state .
teleportation requires quantum correlations , in the form of entanglement , to be shared by the sender and receiver . in this section
we consider how teleportation fidelity changes if decoherence diminishes the extent of the correlation .
we use a decoherence mechanism , phase diffusion , which does not change the intensity ( photon number ) correlations of the entanglement resource , but does destroy the coherence in the number basis .
phase diffusion is modelled by adding random phase fluctuations to each mode independently with the unitary operator @xmath70,\ ] ] where the phase sum and difference @xmath71 is taken to be gaussian randomly distributed with a zero mean and variance @xmath72 , @xmath73 even though a gaussian distribution is not periodic it can be taken to be an approximation of a true periodic distribution , such as @xmath74 , which for sufficiently large @xmath14 , is approximately gaussian near @xmath75 with a variance of @xmath76 .
reference @xcite describes a teleportation protocol using two - mode squeezed vacuum states as an entanglement resource together with number difference and phase sum measurements .
the resource for this protocol is written in the fock basis as @xmath77 the entanglement between resource modes may be altered by changing the squeezing parameter , @xmath78 , and by decohering the resource using phase diffusion . applying the phase shift @xmath79 and averaging over all realisations of the phase gives the resource as a density operator @xmath80 where @xmath81 describes the degree of decoherence .
the number difference measurement can give a positive or negative result and we consider each case separately .
if the state to be teleported is @xmath82 , the output state at the receiver , conditioned on the positive number difference , @xmath36 is @xmath83 with a corresponding fidelity given by @xmath84 where @xmath85 is the probability of obtaining a result @xmath36 for photon number difference measurements at the sender , which does not depend on the decoherence . for measurement of negative number difference , @xmath86 , the fidelity after
teleportation is @xmath87 where @xmath88 the average fidelity as a function of degree of decoherence , @xmath89 , is shown in fig .
[ fig : fsq ] and behaves as we would expect ; decoherence in the resource reduces the output quality of the protocol implying that the entanglement available as a resource for teleportation has decreased . applying our decoherence model to eq .
( [ eq : idealres ] ) and averaging over all realisations of the phase we obtain the total state : @xmath90 where @xmath89 is the degree of decoherence as before .
after completion of the protocol the fidelity is given by @xmath91 where @xmath38 and @xmath92 run from @xmath39 to @xmath14 and @xmath93 is given by eq .
( [ eq : probideal ] ) .
it is not difficult to show that by letting @xmath94 we reproduce the result without noise , eq .
( [ eq : fideal ] ) . the average fidelity as a function of the degree of decoherence , @xmath89 , is shown in fig . [ fig : fidealnoise2 ] for the example of a coherent state , @xmath51 . as the degree of decoherence is increased , the fidelity drops away quickly .
this is because the off - diagonal matrix elements of @xmath95 are being `` washed out '' by the @xmath96 term in the exponential .
physically , we are reducing the entanglement between the resource modes by making measurement of phase more random and would expect the ability of the technique to teleport a state to decrease fig .
[ fig : fidealnoise2 ] shows this effect explicitly .
we add noise to the beam splitter resource state in the same manner as described in sec .
[ sec : sqnoise ] , obtaining the total state , @xmath97 after the teleportation protocol , we find that the fidelity with respect to the initial state is given by @xmath98 as we can see in fig . [
fig : fbsnoise ] , the fidelity decreases due to decoherence in the resource , except that the fidelity decreases from approximately one half instead of one as in sec .
[ sec : ideal ] .
full decoherence corresponds to no entanglement between the resource modes and a completely flat phase probability distribution .
a flat phase probability distribution is equivalent to taking the limit @xmath99 in the fidelities of sec .
[ sec : noise ] thus making the off - diagonal terms in the density matrix representing the output state , @xmath100 , zero .
physically , this limit corresponds to retaining the number correlations but making a measurement of phase completely arbitrary .
we now suggest that this may be considered a _
classical _ limit of the teleportation protocol .
to motivate this point of view , we analyse a classical analogue of the original qubit teleportation protocol@xcite .
consider three classical bits , @xmath101 , where @xmath19 and @xmath20 are correlated bits shared between the sender and receiver respectively .
the bit labelled @xmath62 is the target bit and its state is specified by a distribution , @xmath102 over the values of the binary variable .
the bits @xmath19 and @xmath20 are correlated and have the state @xmath103 where @xmath104 is the usual kronecker delta .
the total state of all three bits is @xmath105 .
we now suppose that the sender can measure the quantity @xmath106 ( addition mod 2 ) on bits @xmath62 and @xmath19 .
the result of this measurement is @xmath107 if both @xmath62 and @xmath19 have the same value and @xmath108 if they have different values .
the sender @xmath19 communicates this result to the receiver @xmath20 .
the conditional state of the receiver given the result of the measurement , @xmath109 is given by standard bayesian conditioning as @xmath110 where @xmath111 is the probability that the joint measurement on @xmath19 and @xmath62 gives the result @xmath109 .
it can be shown that @xmath112 where @xmath113 is the logical not operation .
the receiver @xmath20 knows the result of the joint measurement and can implement a local not operation if the result of the measurement is @xmath108 .
given that local operation , we see that the state of the receiver , @xmath114 is now identical to the state of the target bit , that is to say it has exactly the same probability distribution .
a little thought shows the protocol just described is exactly what would be implemented in the original qubit protocol if the shared resource between @xmath19 and @xmath20 were the completely decohered state @xmath115 .
note that in this case the only information that can be ` teleported ' is the probability distribution for the target bit in the basis in which @xmath95 is diagonal .
for all three entanglement resources considered it can be shown that the average fidelity in the fully decohered limit ( @xmath116 ) reduces to : @xmath117 for example , if the target is a coherent state then this may be shown to be @xmath118 this is the fidelity between a pure state and a totally mixed state with the same photon number distribution . we conclude that if the resource contains only classical intensity correlations it is only possible to teleport the number distribution of the target state : no phase information is teleported . in the sense of the qubit discussion in the previous paragraph , we call this the classical limit of the protocol .
we have shown that a teleportation scheme involving coupled oscillator states using number sum and phase difference measurements is possible , given sufficiently large numbers of fock states in the resource .
the ability of the scheme to reliably teleport a state was shown to improve as the number of fock states in the resource increases . in the case of the beam splitter generated resource this physically means more photons incident on the beam splitter ports .
we have illustrated the effects of decoherence ( in the form of phase diffusion ) in three entanglement resources ( ideal , beam splitter generated and squeezed state ) on the fidelity of teleportation and have related this qualitatively to the change in entanglement of the resource .
the decoherence maintains the classical intensity correlation inherent in the resource . in the limit of complete decoherence
the degraded state is only capable of teleporting the number distribution of the target state . as this result
would have been obtained using standard bayesian conditioning in a classical probabilistic protocol , we argue that it defines a classical limit for the quantum scheme . | we analyse the fidelity of teleportation protocols , as a function of resource entanglement , for three kinds of two mode oscillator states : states with fixed total photon number , number states entangled at a beam splitter , and the two - mode squeezed vacuum state .
we define corresponding teleportation protocols for each case including phase noise to model degraded entanglement of each resource . |
magnetic fields play an important role in the dynamics of many relativistic astrophysical systems such as pulsars , magnetars , gamma - ray burst ( grbs ) and active galactic nuclei ( agns ) . in many of these scenarios , the ohmic diffusion timescales of the magnetized plasma
is much longer than the characteristic dynamical timescale of the system , so one can formally take the limit of infinite electrical conductivity .
this is regarded as the ideal mhd limit , and it is in general a good approximation to describe astrophysical plasmas .
furthermore , such a limit is described by a relatively manageable , but certainly involved hyperbolic system of equations without stiff terms which facilitates its computational implementation .
the ideal mhd limit has been extensively used in the last years to study many of the previous systems ( i.e. , which basically consist of magnetized neutron stars and black hole accretion disks ) in the fully non - linear regime . in spite of its success and convenience , the ideal mhd approximation also has some limitations . at a purely theoretical level
, the assumption of vanishing electrical resistivity prevents some important physical phenomena such as dissipation and reconnection of the magnetic field lines .
reconnection efficiently converts magnetic energy into heat and kinetical energy in very short timescales .
this process is believed to be the mechanism originating many energetic emissions , such as in soft gamma - ray repeaters ( which could be explained by giant magnetar flares ) , the y - point of pulsar magnestosphere or even the short gamma - ray bursts @xcite . in order to describe such processes , schemes going beyond the ideal mhd limit
are required .
at the numerical level , all numerical schemes inherit some numerical resistivity which depends strongly on the resolution , making difficult to disentangle physical phenomena from numerical artifacts especially in highly demanding computational scenarios .
the presence of magnetic fields demands relatively high resolution to accurately capture all the physical processes involved , many of them occurring at very small scales .
this high resolution is particularly important in the case of instabilities which amplify the magnetic field , such as the kelvin - helmholtz instability occurring during the merger of binary neutron stars , and the magneto - rotational instability ( mri ) occurring in accretion disks @xcite .
accurate modeling of the rarefied magnetospheres of compact objects similarly requires high resolution .
the electromagnetic fields in this region may be easier to model by adopting a different limit of the mhd equations known as the _ force - free limit _ @xcite . in this approximation the fluid inertia is neglected , implying that the fluid does not influence directly the dynamics of the electromagnetic fields .
one possibility to overcome these limitations is to consider instead the resistive mhd framework and solve the full maxwell and hydrodynamic equations .
the coupling between these two is provided by the current by a suitable ohm s law. with a convenient choice of current , including both induction and ohmic terms , it is possible to recover both the ideal mhd limit , as well as the finite - resistivity scheme required to describe physical dissipation and reconnections . the effect of small - scales - dynamics can also be modeled with moderate resolutions by using a suitable current .
finally , magnetically dominated magnetospheres can be described by a phenomenological current that decouples the fluid from the force - free em fields .
the numerical evolution of this resistive mhd code is not free of difficulties .
the resistive mhd equations can be regarded as an hyperbolic system with relaxation terms that become stiff for some limits of the current .
consequently , numerical evolution of this system represents a numerical challenge , and several works have recently explored different possibilities to implement it @xcite . in this work
we take a step further in the development of one of these approaches , based on the implicit - explicit ( imex ) runge - kutta .
our aim is to model both the interior and the exterior of a star with a phenomenological current based on physical arguments
. this will be particularly interesting to study the electromagnetic emissions of astrophysical relativistic systems involving magnetized neutron stars .
the capabilities of our approach are tested by considering the force - free aligned rotator solution , a well studied problem in the context of pulsar magnetospheres .
these works were restricted to flat spacetime and excluded the interior of the star from the computational domain , thus side - stepped the stiffness problem mentioned above . recently a hybrid approach , matching both the ideal and force - free system of equations , revisited this problem within a framework capable of studying both star and surrounding magnetosphere within general relativity @xcite .
however such a scheme still relies on two different approximations applied in two regions .
the approach we present here finally allows for treating the system from a global point of view with a single , general relativistic , framework .
we also revisit another important astrophysical scenario with a much less understood dynamics ; the collapse of a magnetized neutron star to a black hole .
this system represents even a more challenging problem because of the strong gravity fields , and it has been studied numerically by considering different approximations .
an early study matched an analytical solution for the star to an electrovacuum magnetosphere @xcite .
more recently , further realism was achieved by adopting the hybrid scheme that matched the numerical solution of the star to force - free magnetosphere @xcite . a step towards studying this system within a common , resistive ,
framework was presented in @xcite , although the star s exterior was treated as an electrovacuum magnetosphere .
our approach presented here is able to consistently study the star and its force - free magnetosphere within the general relativistic resistive mhd equations .
the paper is organized as follows .
section [ sec : equations ] summarizes the fully relativistic resistive mhd system , which is slightly different from the one adopted in @xcite . in section [ sec : coupling_em_fluid ] it is discussed a generic family of algebraic ohm s law , and how to construct a phenomenological current to recover both the ideal mhd and the force - free limits .
section [ sec : hyperbolic_relaxation ] summarizes briefly the imex runge - kutta methods and different techniques to solve generically the implicit step for any algebraic form of the relaxation terms .
the application of these methods to the resistive mhd system is performed is section [ sec : rmhd ] .
section [ sec : simulations ] presents our numerical results for the aligned rotator and the collapse of a neutron star to a black hole .
we conclude with some remarks . throughout this work
we adopt geometric units such that @xmath0 , and the convention where greek indices @xmath1 denote spacetime components ( ie , from @xmath2 to @xmath3 ) , while roman indices @xmath4 denote spatial ones .
bold letters will represent vectors .
this section summarizes the general relativistic resistive magnetohydrodynamic equations , that will allow us to model self - gravitating magnetized fluids .
the evolution of the spacetime geometry is governed by einstein equations .
the electromagnetic fields and the fluid obey , respectively , the maxwell and the general relativistic hydrodynamic equations .
the closure of the system is given by two constitutive equations ; the first one is the equation of state , which relates the pressure to the other fluid variables .
the second one is ohm s law , defining the coupling between the fluid and the electromagnetic fields which will be described in the next section .
the geometry of the spacetime can be obtained by solving the four - dimensional einstein equations .
these equations can be recast as a standard initial value problem by splitting explicitly the time and the space coordinates through a 3 + 1 decomposition , such that the line element can be expressed as @xmath5 where @xmath6 is the spacetime metric , @xmath7 is the intrinsic metric of the spacelike hypersurfaces , and the lapse function @xmath8 and the shift vector @xmath9 relates how the coordinates change between neighboring hypersurfaces .
the normal to the hypersurfaces is given explicitly by @xmath10 indices on spacetime quantities are raised and lowered with the 4-metric and its inverse , while the 3-metric and its inverse are used to raise and lower indices on spatial quantities .
the rate of change of the intrinsic curvature from one hypersurface to another is given by the extrinsic curvature @xmath11 where @xmath12 is the lie derivative along the vector @xmath9 .
at any given time , the spacetime geometry is then fully defined by the @xmath13 variables @xmath14 .
we adopt the baumgarte - shapiro - shibata - nakamura ( bssn ) formulation of einstein s equations to evolve a suitable combination of these fields , in a form very close to the presented in @xcite .
the electromagnetic fields follow maxwell equations , that in their extended version can be written as @xcite @xmath15 where @xmath16 are the maxwell and the faraday tensors , @xmath17 is the electric current and @xmath18 are scalars introduced to control dynamically the constraints by exponentially damping them in a characteristic time @xmath19
when both the electric and magnetic susceptibility of the medium vanish , like in vacuum or in a highly ionized plasma , the faraday tensor is simply the dual of the maxwell one , @xmath20 where @xmath21 is the levi - civita pseudotensor of the spacetime , related to the 4-indices levi - civita symbol @xmath22 by @xmath23 in this case , both tensors can be decomposed in terms of the electric and magnetic fields , @xmath24 such that @xmath25 and @xmath26 are the electric and magnetic fields measured by a normal observer @xmath27 .
both fields are purely spatial , that is , @xmath28 .
the covariant maxwell equations ( [ maxwell_ext_eqs2a ] , [ maxwell_ext_eqs2b ] ) can be written , by performing the 3 + 1 decomposition , in term of the electromagnetic fields and the divergence - cleaning scalars @xcite as , @xmath29 where @xmath30 is the three - dimensional levi - civita pseudotensor .
since @xmath31 is antisymmetric , the four - divergence of equation ( [ maxwell_ext_eqs2a ] ) leads to an additional equation for the current conservation of maxwell solutions , @xmath32 the electric current @xmath17 can be decomposed into components along and perpendicular to the vector @xmath33 , @xmath34 where @xmath35 and @xmath36 are the charge density and the current as observed by a normal observer @xmath33 .
again , @xmath36 is purely spatial , so @xmath37
. the current conservation ( [ conserved_current ] ) can be expressed , with the 3 + 1 decomposition , as @xmath38 only a prescription for the spatial components @xmath39 , which will determine the coupling between the em fields and the fluid , is required to complete the system of maxwell equations .
this relation , commonly known as ohm s law , will be discussed in detail in section [ sec : coupling_em_fluid ] . a perfect fluid minimally coupled to an electromagnetic field is described by the total stress - energy tensor @xmath40 u_{\mu } u_{\nu } + p g_{\mu \nu } \nonumber \\ & + & { f_{\mu}}^{\lambda } f_{\nu \lambda } - \frac{1}{4 } g_{\mu \nu } ~ f^{\lambda \alpha } f_{\lambda \alpha } \end{aligned}\ ] ] where a factor @xmath41 has been absorbed in the definition of the electromagnetic fields . here
@xmath42 is the rest mass density , @xmath43 the internal energy and @xmath44 is the pressure , given by a closure relation @xmath45 commonly known as the equation of state ( eos ) .
these fluid quantities are measured in the rest frame of the fluid element .
however , to describe the system is usually more convenient to adopt an eulerian perspective where coordinates are not tied to the flow of the fluid .
the four - velocity @xmath46 describes how the fluid moves with respect to the eulerian observers , and can be decomposed into space and time components , @xmath47 where @xmath48 corresponds to the familiar three - dimensional velocities as measured by eulerian observers ( i.e. , @xmath49 ) .
the time component is defined by the normalization relation @xmath50 , such that @xmath51 where we can now recognize @xmath52 as the lorentz factor .
in summary , the magnetized fluid is described by the physical fields ( i.e. , the fluid variables and the electromagnetic fields ) plus the divergence cleaning scalars , which form the set of primitive variables @xmath53 the matter evolution must comply with the conservation of the total stress - energy tensor @xmath54 which can be expressed as a system of conservation laws for the energy density @xmath55 and the momentum density @xmath56 , defined from the projections of the stress - energy tensor @xmath57 in addition to the conservation of energy and momentum , the fluid usually also conserves the total number of particles , @xmath58 where @xmath59 is the baryon number density .
this equation is just the relativistic generalization of the conservation of mass . as mentioned above ,
it is necessary to specify the eos to define the pressure and complete the system of hydrodynamic equations . along this paper
we will consider either the polytropic eos @xmath60 , which is a good approximation to describe cold stars , and the ideal gas eos @xmath61 , which allows for shock heating in the fluid .
the evolution of the electromagnetic fields follows the maxwell equations and the conservation of charge , while the fluid fields are governed by the conservation of the total energy , momentum and baryonic number . in order to capture accurately the weak solutions of these non - linear equations in presence of shocks
it is important to express them as a set of local conservation laws , namely @xmath62 \\ & = & - \sqrt{\gamma } b^k ( \partial_k \beta^i ) + \sqrt{\gamma } \phi \left ( \gamma^{ij } \partial_j \alpha - \alpha \gamma^{jk } \gamma^i_{jk } \right ) \nonumber \\ \label{maxwell12 } \partial_t ( \sqrt{\gamma } e^i ) & + & \partial_k [ \sqrt{\gamma } \left ( -\beta^k e^i - \alpha ( \epsilon^{ikj } b_j - \gamma^{ik } \psi ) \right ) ] \\ & = & - \sqrt{\gamma } e^k ( \partial_k \beta^i ) + \sqrt{\gamma } \psi \left ( \gamma^{ij } \partial_j \alpha - \alpha \gamma^{jk } \gamma^i_{jk } \right ) \nonumber \\ & & - \alpha \sqrt{\gamma } j^i \nonumber \\
\label{maxwell13 } \partial_t ( \sqrt{\gamma } \phi ) & + & \partial_k [ \sqrt{\gamma } ( - \beta^k \phi + \alpha b^k ) ] \\ & = & \sqrt{\gamma } [ - \alpha\ , \phi\ , trk + b^k ( \partial_k \alpha ) -\alpha \kappa \phi ] \nonumber \\
\label{maxwell14 } \partial_t ( \sqrt{\gamma } \psi ) & + & \partial_k [ \sqrt{\gamma } ( - \beta^k \psi + \alpha e^k ) ] \\ & = & \sqrt{\gamma}[- \alpha\ , \psi\ , trk + e^k ( \partial_k \alpha ) + \alpha q -\alpha \kappa \psi ] \nonumber \\
\label{maxwell15 } \partial_t ( \sqrt{\gamma } q ) & + & \partial_k [ \sqrt{\gamma } ( - \beta^k q + \alpha j^k ) ] = 0 \\ \label{fluid11 } \partial_t ( \sqrt{\gamma } d ) & + & \partial_k [ \sqrt{\gamma } ( - \beta^k + \alpha v^k ) d ] = 0 \\ \label{fluid12 } \partial_t ( \sqrt{\gamma } \tau ) & + & \partial_k [ \sqrt{\gamma } \left ( - \beta^k \tau + \alpha ( s^k - v^k d ) \right ) ] \\ & = & \sqrt{\gamma } [ \alpha s^{ij } k_{ij } - s^j \partial_j \alpha ] \\
\label{fluid13 } \partial_t ( \sqrt{\gamma } s_i ) & + & \partial_k [ \sqrt{\gamma } ( - \beta^k s_i + \alpha { s^k}_i ) ] \\ & = & \sqrt{\gamma } [ \frac{\alpha}{2 } s^{jk } { \partial_{i } } \gamma_{jk } + s_j \partial_i \beta^j - ( \tau + d ) \partial_i \alpha ] \nonumber\end{aligned}\ ] ] where we have defined @xmath63 and the enthalpy @xmath64 .
this form of the relativistic resistive mhd equations is basically the same presented already in @xcite .
another similar formulation has also been derived recently @xcite .
notice also that the energy conservation has been expressed in terms of the quantity @xmath65 to recover the newtonian limit of the energy density .
maxwell and hydrodynamic equations are coupled by means of the current @xmath66 , whose explicit form generically depends on the electromagnetic fields and the local fluid properties measured in the comoving frame . consequently , it is convenient to introduce the electric and magnetic fields measured by an observer comoving with the fluid , namely @xmath67 and @xmath68 .
notice that , since @xmath69 , there are only three independent components . the maxwell and faraday tensors can therefore be expressed as @xmath70 and the electric current can be decomposed into components along and transverse to @xmath71 , @xmath72 where @xmath73 and @xmath74 is the charge density measured by the comoving observer .
the relation with the eulerian quantities ( [ current_decomposition ] ) can be obtained from @xmath75 substituting these results into eq .
( [ current_decomposition2 ] ) and using the 3 + 1 decomposition of the four - velocity , one can write the spatial components of the current as @xmath76 since the charge density follows directly from the current conservation ( [ consj_3 + 1 ] ) , the prescription for the three - dimensional electrical current @xmath77 is the only missing piece to completely determine maxwell equations .
a standard prescription , known as the ohm s law , is to consider that the current is proportional to the lorentz force acting on a charged particle , implying a linear relation between the current and the electric field in the comoving frame .
a richer variety of physical phenomena may be described by including also additional terms proportional to the comoving magnetic field , leading to a generalized covariant ohm s law of the form , @xmath78 being @xmath79 the electrical conductivity of the medium @xcite and @xmath80 a parameter related to the covariant generalization of the mean - field dynamo @xcite .
the electrical conductivity can be calculated either in the collision - time approximation @xcite or in the framework of relativistic charged multifluids @xcite , leading to the same main results .
the tensorial conductivity can be written as , @xmath81 where the coefficients are given by @xmath82 here @xmath83 is the collision or relaxation time , @xmath84 is the electron density and @xmath85 and @xmath86 are the electron s charge and mass . in the framework described in @xcite ,
@xmath87 is introduced as a proportionality constant in the dissipative force between the two components of the fluid .
it is easy to check that the first term of the conductivity ( [ tensorial_conductivity ] ) leads to the well known isotropic scalar case , while the other two represent the anisotropies due to the presence of a magnetic field , corresponding to the hall effect . in order to compute the closure relation ( [ current_decomposition3 ] ) it is necessary to write the general relativistic ohm s law in terms of fields measured by an eulerian observer .
let us first consider a simplified ohm s law neglecting both the dynamo effects and the last term in the tensorial conductivity ( [ tensorial_conductivity ] ) , @xmath88~~,\ ] ] as it has also been used in @xcite .
it was pointed out that this current implies an incomplete hall effect @xcite , but it will be enough for our later discussion . within these assumptions , and using that the electric and magnetic fields in the fluid frame can be written as @xmath89 it is straightforward to obtain the contraction @xmath90 \\ & = & \frac{\sigma}{1 + \xi^2 b^2 } [ -w ( e^k v_k ) - w \xi^2 ( e^j b_k ) ( b^k v_k ) ] ~~. \nonumber\end{aligned}\ ] ] the prescription for the spatial current ( [ current_decomposition3 ] ) can be now computed , leading to @xmath91\ ] ] where we have introduced the shortcuts @xmath92 ~,~~ \\ \label{newb } { \cal b}_i & = & w \left [ b_i - \epsilon_{i j k } v^{j } e^{k } - ( v_{k } b^{k } ) v_{i } \right ] ~.\end{aligned}\ ] ] it is important to recall that this current accounts not only for isotropic resistivity but also for some anisotropic effects induced by the magnetic fields . in the regime of low magnetization ( i.e. , @xmath93 ) these anisotropic effects are expected to be small , implying @xmath94 . in this limit
the third term in the current ( [ ohm_relativistic_spatial ] ) can be neglected , leading to the well - known isotropic ohm s law .
the high conductivity of the fluid implies that , in order to get a finite current , the electric field measured by the comoving observers must vanish @xmath95 this is the ideal - mhd condition , which states that the electric field is not an independent variable since it can be obtained via a simple algebraic relation from the velocity and the magnetic vector fields .
the anisotropic effects are expected to be important in magnetically dominated fluids ( i.e. , @xmath96 ) . in this limit @xmath97 , and the second term in the current ( [ ohm_relativistic_spatial ] ) can be neglected . in highly conducting fluids a finite current
is recovered only if the electric field is perpendicular to the magnetic field , @xmath98 since the initial assumption of magnetically dominated fluid prevents the trivial solution @xmath99 . in the next subsection
it will be shown that this relation is one of the constraints of the force - free approximation .
the magnetospheres of magnetized neutron stars @xcite and black holes immersed in externally sourced magnetic fields @xcite are filled with a low - density plasma so rarefied that even moderate magnetic fields stresses can easily dominate over the pressure gradients . in this regime , the main contribution to the stress - energy tensor comes from the electromagnetic part , @xmath100 .
allowing by maxwell equations , the total conservation of energy and momentum can be written as @xmath101 the vanishing of the lorentz force @xmath102 leads to an approximation known as force - free limit , which is valid only for magnetically dominated plasmas with negligible inertia .
the spatial components of the force - free condition ( [ div - stress - fluidem2 ] ) , after performing the 3 + 1 decomposition , are @xmath103 or , after some simple manipulations , @xmath104 where we have defined @xmath105 as the drift velocity .
several options have been proposed to compute the term @xmath106 , which is crucial to provide a completely explicit relation for the current .
for instance , a closed formed for the current can be calculated by enforcing the constraint @xmath107 @xcite .
another option is to evolve maxwell equations by considering only the drift term of the current ( [ forcefree3 ] ) , and correct the electric field after each timestep to satisfy the other force - free condition @xmath108
this approximation has been used successfully to study numerically pulsar magnetospheres @xcite and jets emerging from black holes with an externally sourced magnetic field @xcite .
the numerical evolution of the ideal mhd equations typically fails in low density regions with high magnetization unless sufficient resolution is available , a situation that arises commonly in the magnetospheres .
a standard practice to avoid these failures is to maintain a density floor ( i.e. , the so called _ atmosphere _ ) in regions of low density to exploit advanced numerical techniques for relativistic hydrodynamics .
the density in the atmosphere is much smaller than that inside the star , so this approach does not affect the star s dynamics .
however , in the magnetosphere the fluid inertia ( and pressure ) is typically much smaller than that of the electromagnetic field and one generally encounters numerical difficulties .
these problems are mitigated by increasing the density in the atmosphere , effectively decreasing the magnetization in the exterior of the star .
although these modifications produce an unphysical modeling of the plasma in the magnetosphere , one could still solve correctly maxwell equations by using a suitable current that decouples the electromagnetic fields from the fluid variables .
as explained earlier , the covariant current ( [ ohm_relativistic_spatial ] ) reduces to the ideal mhd limit for high isotropic conductivities ( i.e. , @xmath113 and @xmath114 ) , while that the force - free constraint @xmath108 is enforced for large anisotropic conductivities ( i.e. , @xmath115 ) .
this suggests that the solutions for the em fields in both limits can be achieved just by changing the anisotropic conductivity , independently on the plasma magnetization .
although ohm s law ( [ ohm_relativistic_spatial ] ) is quite general , it still couples the em fields to the velocity .
in addition , the parameter @xmath116 is not appropriate to model the fast decay of the magnetic field with the distance to the source . to overcome these difficulties , and
in part motivated by the strategy introduced in @xcite , we introduce the following phenomenological current to include both the ideal mhd and the force - free limits , @xmath117 \\ & + & \frac{\sigma}{1 + \zeta^2 } \left[{\cal e}^i + \frac{\zeta^2}{b^2 } \ { ( e^k b_k ) b^i + \chi ( e^2-b^2 ) e^i\ } \right]~ , \nonumber\end{aligned}\ ] ] where @xmath118 is a function which vanishes whereas the ideal mhd limit is valid , and tends to @xmath119 whereas the force - free limit is more appropriate .
the anisotropic ratio @xmath120 , which can be reinterpreted from the definition @xmath121 , can be conveniently set to be a constant in the region where the force - free limit is valid .
the physical condition @xmath122 is enforced through a new current term proportional to an anomalous conductivity @xmath123 , which only appears whenever @xmath124 .
overdamping of the electric field is avoided by setting this anomalous conductivity to the characteristic decay time @xmath125 , which can be estimated from the time evolution of @xmath126 .
let us consider the particular astrophysical scenario of magnetized neutron stars .
the large fluid conductivity , both inside and outside the star , is modeled by using a large constant @xmath127 .
the anisotropic ratio , which defines the regions described either with the ideal mhd or the force - free limits , is defined as @xmath128 .
this choice ensures that the interior of the star ( i.e. , @xmath129 ) is dominated by a large isotropic conductivity , reducing the system of equations to the ideal mhd limit .
the exterior of the star ( i.e. , @xmath130 ) is dominated by the anisotropic terms which enforce the force - free condition .
the kernel function @xmath118 is defined such that vanishes inside the star and its value becomes unity outside .
a smooth transition between the inner and the outer region is achieved by using @xmath131 we typically adopt @xmath132 and @xmath133 , being @xmath134 the value for the density of the magnetosphere .
the general system of relativistic resistive mhd equations ( [ maxwell11]-[fluid13],[ohm_ideal_forcefree ] ) brings about a delicate issue when the conductivity in the plasma undergoes very large spatial variations . in regions with high conductivity ,
in fact , the system will evolve on timescales which are very different from those in the low - conductivity region .
mathematically , therefore , the problem can be regarded as a hyperbolic system with relaxation terms which requires special care to capture the dynamics in a stable and accurate manner .
the prototype of these systems can be written as @xmath135 where @xmath136 is the relaxation time . in the limit @xmath137
the system is hyperbolic with spectral radius @xmath138 ( i.e. , the absolute value of the maximum eigenvalue ) . in the other limit @xmath139
the system is clearly stiff since the time scale of the relaxation ( or stiff term ) @xmath140 is much smaller than the maximum speed @xmath138 of the hyperbolic part @xmath141 . in the stiff limit ( @xmath139 ) the stability of an explicit time evolution scheme
is only achieved with a time step size @xmath142 , a much stronger restriction than the cfl condition @xmath143 of the hyperbolic systems .
the development of stable and efficient numerical schemes to overcome this restrictive constraint is challenging , since in many applications the relaxation time can vary many orders of magnitude .
different alternatives to deal with the inherent stiffness of the relativistic resistive mhd equations has been proposed in the last decade ; combination of splitting methods and analytical solutions @xcite , discontinuous galerkin methods @xcite and implicit - explicit ( imex ) runge - kutta methods @xcite .
the following subsections summarize the imex runge - kutta schemes , a family of time integrators which are able to deal with the potentially stiffness issues and are relatively easy to incorporate into an existing relativistic ideal mhd code . an efficient way to solve the hyperbolic - relaxation systems is based on the imex runge - kutta methods . within this scheme , all the fields
are evolved by using a standard explicit time integration except the potentially stiff terms , which are evolved with an implicit time discretization . for the generic system ( [ stiff_equation ] ) this scheme takes the form @xcite @xmath144 where @xmath145 are the auxiliary intermediate values of the runge - kutta .
the coefficients can be represented as @xmath146 matrices @xmath147 and @xmath148 such that the resulting scheme is explicit in @xmath149 ( i.e.,@xmath150 for @xmath151 ) and implicit in @xmath87 .
an imex runge - kutta is characterized by these two matrices and the coefficient vectors @xmath152 and @xmath153 .
notice that at each substep the auxiliary intermediate values @xmath145 involves solving an implicit equation .
since the simplicity and efficiency of solving the implicit part at each step is of great importance , it is natural to consider diagonally implicit runge - kutta ( dirk ) schemes ( @xmath154 for @xmath155 ) for the stiff terms .
a deeper discussion on the imex schemes and the detailed form of the schemes considered here are presented in appendix [ appendixa ] .
the vector of evolved fields @xmath156 can be split in two sets of variables @xmath157 , depending on whether or not they contain any relaxation term in their evolution equations .
the evolution system can then be generically written as @xmath158 where we have considered that the relaxation parameter @xmath43 can be any function not depending directly on the present value of the @xmath159-fields .
the procedure to compute each auxiliary step @xmath145 can be split in two stages : 1 .
compute first the intermediate values @xmath160 which involves information from previous steps , @xmath161 2 .
include the relaxation term at the present time by solving the implicit equation @xmath162 which clearly involves only the @xmath159-fields .
the complexity of inverting this implicit equation depends on the particular form of the relaxation terms . from now on we will restrict ourselves to the algebraic case @xmath163 . next it is described two different ways to solve this implicit equation ; the first one can only be applied when @xmath164 is a linear function , whereas the second one allows @xmath164 to have any non - linear dependence .
the simplest case , however enough to cover a broad range of interesting situations , is to consider a linear relaxation term @xmath166 the implicit equation ( [ second_step ] ) can then be trivially solved @xmath167 \nonumber \\
m & = & [ i - a_{ii}~\frac{\delta t } { \epsilon^{(i ) } } a({\bf w^{(i)}})]^{-1 } ~.\end{aligned}\ ] ] the matrix inversion can be performed analytically and written in a compact form for most of the interesting cases , so that the implicit step can be solved in a completely explicit way . in the more general case
with an arbitrary non - linear dependence it is usually not feasible to solve analytically the implicit step , requiring some approximation to find the solution . a convenient approach to solve
this problem is to linearize the stiff term around an approximate solution @xmath168 , such that @xmath169 notice that we are linearizing around the solution @xmath170 , which is already known at the beginning of the implicit step . by defining @xmath171 , and substituting the previous expansion ( [ linearization2 ] ) in ( [ second_step ] ) ,
it is obtained @xmath172\ ] ] this implicit equation can be written , after some manipulations , in the following way @xmath173 \nonumber \\ m & \equiv & [ i - a_{ii}~\frac{\delta t } { \epsilon^{(i ) } } a({\bf \bar{v}},{\bf w}^{(i)})]^{-1}\end{aligned}\ ] ] the final expression ( [ invert_matrix_linearized _ final ] ) can be solved through a newton - raphson iterative procedure such that , at each iteration @xmath174 , uses an initial guess @xmath175 to find the next approximate solution @xmath176 .
we adopt finite difference techniques on a regular cartesian grid to solve the problems of interest . to ensure sufficient resolution
is achieved in an efficient manner we employ adaptive mesh refinement ( amr ) via the had computational infrastructure that provides distributed , berger - oliger style amr @xcite with full sub - cycling in time , together with an improved treatment of artificial boundaries @xcite .
the refinement regions are determined using truncation error estimation provided by a shadow hierarchy @xcite which adapts dynamically to ensure the estimated error is bounded within a pre - specified tolerance .
the spatial discretization of the geometry is performed using a fourth order accurate scheme , while that high resolution shock capturing methods based on the hlle flux formula with ppm reconstruction are used to discretize the resistive mhd variables @xcite .
the time - evolution is performed through the method of lines using a third order accurate implicit - explicit runge - kutta integration scheme described in the previous section .
we adopt a courant parameter of @xmath177 so that @xmath178 on each refinement level @xmath179 . on each level
, one therefore ensures that the courant - friedrichs - levy ( cfl ) condition dictated by the principal part of the equations is satisfied .
the relaxation terms of the resistive mhd system are associated to the current , which mainly appears in the time evolution equation of the electric field .
the evolved fields can then be split into and non - stiff @xmath180 and potentially stiff @xmath181 .
the evolution of the non - stiff fields is performed by the explicit part of the imex runge - kutta , and it is very similar to a standard implementation of the ideal mhd equations .
the evolution of the electric field contains in addition the relaxation terms , namely @xmath182 ~ , \nonumber \\ & & - \sqrt{\gamma } e^k ( \partial_k \beta^i ) + \sqrt{\gamma } \psi \left ( \gamma^{ij } \partial_j \alpha - \alpha \gamma^{jk } \gamma^i_{jk } \right ) \nonumber \\ & & - \alpha \sqrt{\gamma } j^i_{e } ~ , \nonumber\\ r_e & = & -\alpha j^i_{s } ~. \nonumber\end{aligned}\ ] ] where the factor @xmath183 , corresponding to the fluid conductivity , is absorbed in the definition of @xmath184 .
the current has been split into a potentially stiff part , @xmath185 , and the terms which can be treated explicitly , @xmath186 . for the phenomenological ohm s law ( [ ohm_ideal_forcefree ] ) these components can be written explicitly as @xmath187 ~ , \\
j^i_{s } & = & \frac{\sigma}{1 + \zeta^2 } \left[{\cal e}^i + \frac{\zeta^2}{b^2 } \ { ( e^k b_k ) b^i + \chi ( e^2-b^2 ) e^i\ } \right]~. \nonumber\end{aligned}\ ] ] notice that , although the evolution of @xmath35 is driven by the current , these terms do not become potentially stiff in this equation since they are not proportional to the field itself .
however , the delicate balance between the different fields in the current , which allows to get finite values even for very high conductivities , may be broken during the reconstruction of the fields at the interfaces .
these unacceptable large errors are prevented in the standard implementations of the force - free equations by computing the charge density from the constraint @xmath188 instead of using the charge conservation .
the resulting set of equations is still hyperbolic , since the charge density only couples to the em fields throughout the non - principal term @xmath189 @xcite .
here we prefer to keep the charge density as an evolution field and treat all the fields in the same manner .
the errors at the interfaces are avoided by performing directly the reconstruction of the current @xmath39 , which is computed just after solving the stiff terms .
this ensures that the fluxes of @xmath35 will remain bounded between the values given by well - defined neighboring points .
the numerical evolution of the resistive mhd system ( [ maxwell11]-[fluid13 ] ) involves the recovery , after each timestep , of the primitive fields @xmath190 from the conserved or evolved fields @xmath191 .
although the conserved fields are just algebraic relations of the primitive ones , the opposite is not true ; due to the enthalpy and the lorentz factor these quantities are related by complicated equations that can only be solved numerically , except for particularly simple equations of state .
the solution at time @xmath192 is directly obtained , for most of the conserved quantities , by evolving their ( non - stiff ) evolution equations .
however , the explicit evolution of the potentially stiff fields only provides a partial solution .
as explained in the previous section , a complete solution for the electric field involves taking into account the relaxation terms by solving the corresponding implicit equation . for a generic ohm s law
, these relaxation terms will depend on the velocity and other primitive fields .
nevertheless , the recovery of the primitive variables from the conserved ones involves all the fields , including the electric field .
this is a consistency constraint which implies that the recovery process and the implicit step evolution must be solved _ at the same time_. we will next describe an iterative procedure to evolve the stiff part and recover the primitive fields for the phenomenological current ( [ ohm_ideal_forcefree ] ) , as described in subsection [ subsubsec : nonlinearr ] . 1 . to start the iterative process it is required an approximate solution initial guess for the electric field @xmath193 and the fluid unknowns of the system , that we have chosen to be the single combination @xmath194 .
the initial guess for this unknown is given simply by the previous time step @xmath195 .
possible choices for the electric field initial guess are : * the previous time step @xmath196 * the ideal mhd limit @xmath197 , which involves performing first the recovery in the ideal mhd case ( see appendix b for details ) . * the approximate solution given by the explicit and previous implicit step evolutions @xmath198 . * the trivial case @xmath199 .
+ it may be difficult to estimate a priori which initial guess is more convenient .
for this reason , our scheme starts with the first option and , if no solution is found , tries sequentially the other choices .
2 . subtract the electromagnetic contributions from the energy and momentum densities , @xmath200 such that the lorentz factor can be computed as , @xmath201 3 .
write also the pressure as a function of the conserved variables and the unknown @xmath202 .
for the ideal gas eos @xmath61 this relation is just @xmath203 4 .
obtain an equation @xmath204 , written in terms of the unknown @xmath202 and the conserved fields , such that it is satisfied only for true solutions of @xmath202 . by using the previous expression ( [ c2p_pressure ] ) in the definition of @xmath205 ,
we can write @xmath206 x + [ \frac{\gamma-1}{\gamma w } -1 ] d - { \tilde \tau } \,,\ ] ] where @xmath52 is computed through eq.([con2prim_2 ] ) .
the equation @xmath204 can be solved numerically by using an iterative newton - raphson solver .
the solution in the iteration @xmath207 can be computed as @xmath208 the derivative of the function @xmath209 can be computed analytically , @xmath210 5 . update the primitive fields by using the relations @xmath211 6 .
update the electric field with the updated values of the primitive fields by solving the implicit equation , corresponding to eq .
( [ second_step ] ) , + @xmath212 + which can be formally solved with the method described in subsection [ subsubsec : nonlinearr ] for @xmath213 , that is , @xmath214 \\
m & = & [ i - a_{ii } \delta_t a]^{-1}~~,~~ a = \frac{\partial r_e^i}{\partial e^j } ~~. \end{aligned}\ ] ] + for the phenomenological ohm s law ( [ ohm_ideal_forcefree ] ) , the matrix @xmath215 to be inverted is @xmath216 \right.\ } \biggr ] \nonumber \end{aligned}\ ] ] with @xmath217 .
iterate until the solution @xmath218 satisfies their constitutive equations @xmath219 , being @xmath220 defined by equation ( [ invert_matrix_linearized5 ] ) . in occasions
the recovery procedure is unable to find a physical state for a given set of conserved variables . in such cases , which usually occur near a star s surface
, failures can be avoided by assuming that the fluid is isentropic in that timestep and therefore satisfying a polytropic eos @xmath221 . since
the internal energy is also a function of the density ( i.e. , @xmath222 ) for isentropic processes , the conserved quantities are overdetermined and the energy equation can be neglected in the recovery procedure , leading to a more robust algorithm .
notice also that , although our discussion was focused on the phenomenological the ohm s law ( [ ohm_ideal_forcefree ] ) , the method described in subsection [ subsubsec : nonlinearr ] can be applied to any algebraic form of the current . even more general cases with derivative terms can be considered , with the condition that those must be evaluated at earlier times . in a similar way
, the method for linear relaxation terms described in subsection [ subsubsec : linearr ] can be generically used for non - linear algebraic currents with the condition that the non - linear terms are evaluated at previous time steps , as it was considered in @xcite .
this option does not require an initial guess for the electric field and therefore may be more effective in avoiding unphysical states .
in this section we report our numerical studies of astrophysical scenarios involving the dynamical evolution of a rotating magnetized star and its magnetosphere .
the initial data of rigidly rotating neutron stars is provided by the lorene package _ magstar _ , which adopts a polytropic equation of state @xmath223 with @xmath224 , rescaled to @xmath225 .
because the fluid pressure in a neutron star is many orders of magnitude larger than the electromagnetic one , moderate magnetic fields will have an insignificant effect on both the geometry and the fluid structure , and so they can be specified freely . for this reason we have chosen an initial poloidal magnetic field inside the star that becomes dipolar in the external region .
the electric fields are set by assuming the ideal mhd condition , with an initial zero fluid velocity in the magnetosphere . during the evolution , which is performed with the methods described in the previous sections , the ideal mhd and
the force - free limits are enforced inside / outside the star by using the phenomenological current ( [ ohm_ideal_forcefree ] ) .
we monitor the electromagnetic luminosity , constructed from the newman - penrose scalar @xmath226 @xcite , @xmath227 that accounts for the energy carried off by outgoing waves to infinity and it is equivalent to the poynting luminosity at large distances .
additionally we monitor the ratio of particular components of the maxwell tensor @xmath228 which , in the stationary , axisymmetric case , can be interpreted as the rotation frequency of the electromagnetic field @xcite .
we consider first the evolution of an uniformly rotating stable star of mass @xmath229 and equatorial / polar radius @xmath230 .
the star rotates with a period @xmath231 , so that the light cylinder is located at @xmath232 .
the strength of the magnetic field at the pole is @xmath233 .
the numerical domain extends up to @xmath234 and contains four centered fmr grids with decreasing sizes ( and twice better resolved ) such that the highest resolution grid has @xmath235 and extends up to @xmath236 ( i.e. , beyond the light cylinder ) .
this initial configuration is evolved until that the solution relaxes to a quasi - stationary state .
different quantities are plotted along the equatorial plane in fig .
[ fig : pulsar1d ] and that both the initial and the final magnetic field solutions are displayed in fig .
[ fig : pulsar_bfield ] .
the relaxed final state has the characteristic features observed in previous works . the magnetic fields are being dragged by the fluid rotation in the interior of the star ( i.e. , as in the initial state ) , producing a tension that forces the magnetic fields in the magnetosphere to co - rotate with the star up to the light cylinder . beyond this surface ,
the magnetic field lines open up , creating a current sheet in the equatorial plane where the anomalous resistivity in the current ( or bringing back the neglected fluid inertia ) is necessary to preserve the physical condition @xmath237 .
we have computed the poynting - vector luminosity at two surfaces at @xmath238 located outside the light cylinder , where the measures converge to a unique well - defined value .
the em radiation is mainly dipolar ( i.e. , around @xmath239 of the energy ) , with a small fraction in higher multipoles .
the luminosity can be compared with previous results in flat spacetime geometry where the spherical star is modeled through inner boundary conditions @xcite @xmath240 our results agrees within a difference of @xmath241 , where we have used @xmath242 .
it is unclear where this small disagreement may come from , since there are several possible explanations ; the ambiguity in the definition of the radius of oblated stars , an excess of dissipation in the current sheet , or purely strong gravitational effects , which may become important due to the high compactness @xmath243 of the star .
we have also monitored both the energy - momentum constraints and the divergence constraints , checking that they remain small and under control during the evolution .
in particular , @xmath244 in all the domain but the current sheet . by comparing the solutions obtained with three different resolutions , each one improving a factor @xmath245 the previous space discretization @xmath246 ,
we have observed that the code converges at @xmath247-order .
the luminosity for these three resolutions displayed in fig .
[ fig : convergence_luminosity ] shows that , in spite of the spasmodic reconnections happening in the current sheet , the system converges to a quasi - stationary solution with a steady luminosity .
+ after assessing the validity of our implementation with the aligned rotator solution , we can consider a more challenging and dynamical case ; the collapse of an uniformly rotating magnetized neutron star to a black hole .
the initial data is the same as it was considered in @xcite ; a star lying on the unstable branch with mass @xmath248 and equatorial / polar radius @xmath249 , rotating with a period @xmath250 so that the light cylinder is located at @xmath251 .
the strength of the magnetic field at the pole is chosen to be @xmath252 , although the results may be rescaled to any strength as long as the magnetic pressure is much smaller than the fluid one .
the numerical domain extends up to @xmath234 and contains @xmath253 centered fmr grids with decreasing sizes such that the highest resolution grid has @xmath254 and extends up to @xmath255 , while that the second highest extends up to @xmath256 , beyond the initial location of the light cylinder .
small perturbations arising from numerical truncation errors are enough to trigger the collapse of the unstable star .
the horizon appears after around @xmath257 , although the most dynamical part only stands for the last @xmath258 , ending when all the matter disappears beyond the horizon and the nearby magnetic fields reconnects in the equatorial plane and escapes to infinity .
the conservation of angular momentum implies that the angular velocity of the star increases during the collapse , dragging the magnetic field lines in the magnetosphere and bringing the light cylinder closer to the star .
the magnetic fields also grow due to the magnetic flux conservation .
once all the fluid has accreted onto the black hole , the magnetic fields looses their anchorage , reconnects and propagates away from the source .
a significant fraction of the energy stored in the magnetosphere is radiated to infinity in this burst .
the density of the star , the poynting vector density @xmath259 and the magnetic fields are displayed at some representative stages of the collapse in fig .
[ fig : collapse_bfield ] .
the growth of the angular velocity and the magnetic field implies that the luminosity of the aligned rotator ( [ eq : lsd_dipole ] ) during a quasi - adiabatic collapse will increase as @xmath260 @xcite , being @xmath261 the initial luminosity of the star .
however , since the collapse time is shorter than the star s period , the outer part of the magnetosphere is not able to respond to the changes in the start s surface , reducing the power of the luminosity to @xmath262 @xcite .
in addition , strong gravitational effects will soften the growth of both the angular frequency and the radial magnetic field , leading to a much more moderate luminosity growth .
we have computed the electromagnetic luminosity in a sphere located at @xmath263 , beyond the light cylinder .
the em radiation is mainly dipolar and grows during the collapse , with a strong burst due to the reconnection when the fluid is completely swallowed by the black hole .
the luminosity and the angular velocity computed inside and outside the star are displayed in figure [ fig : collapse_luminosity ] .
the energy in the magnetosphere increases by a factor @xmath264 during the collapse .
the total radiated energy can be expressed as a fraction @xmath265 of the peak energy @xmath266 , namely @xmath267 where we have used @xmath268 for a star of radius @xmath269 @xcite . in our simulation
we have found @xmath270 , implying that the system radiates @xmath271ergs during the collapse ( for a magnetic field of @xmath272 g ) .
notice that this value is different from the analytical estimates and indicates the importance of the fast dynamic and strong gravitational effects in this scenario .
we have presented a formulation of the general relativistic resistive mhd equations . we have discussed different generalizations of the isotropic ohm s law , and constructed a phenomenological current such that the system reduces either to the ideal mhd limit or to the force - free approximation just by changing the ratio of isotropic / anisotropic conductivities .
we have explained how to deal with the potential stiffness of the equations by using the implicit - explicit runge - kutta methods , showing how to perform the implicit evolution of the electric field and the recovery of the primitive from the conserved fields at the same time for any algebraic ohm s law .
we implemented the formulation within the had computational infrastructure and revisited two interesting astrophysical problems ; the aligned rotator and the collapse of a rotating neutron star to a black hole .
none of these cases has a known analytical solution , although the first case has been studied extensively .
we find a reasonable agreement between our results and previous studies of the aligned rotator , recovering the same qualitative features and approximately the same electromagnetic luminosity .
the case of the collapsing star is more challenging and has been only studied previously either assuming an electrovacuum magnetosphere and/or by matching the exterior to the interior solution .
our results are qualitatively similar to those found in @xcite , although the total radiated energy in our simulations is one order of magnitude larger due to an increase in both the peak energy in the magnetosphere and the fraction of radiated energy .
the possible detectability of this burst has been already discussed in detail in @xcite and therefore will not be repeated here . in conclusion
, the resistive mhd framework allows to consider a broad range of new phenomena;study reconnections and dissipation with more realistic ohm s law - like the resistive solutions of pulsar magnetospheres @xcite- , model the magnetic growth due to different instabilities by using the mean - field dynamo @xcite , and compute the magnetosphere interaction of binary systems like neutron - neutron stars and neutron - black hole , which may be crucial to study the possible electromagnetic counterparts to the gravitational waves emitted by these systems , among others possibilities .
work on these directions is in progress and it will be reported in the near future .
imex runge - kutta schemes can be represented by a double tableau in the usual butcher notation @xcite @xmath273 where the coefficients @xmath274 and @xmath275 used for the treatment of non - autonomous systems are given by the following relation @xmath276 solutions of conservation equations have some norm that decreases in time .
it would be desirable , in order to avoid spurious numerical oscillations arising near discontinuities of the solution , to maintain such property at a discrete level by the numerical method .
the most commonly used norms are the tv - norm and the infinity norm .
a scheme is called strong stability preserving ( ssp ) if maintains a given norm during the evolution @xcite . in all these schemes
the implicit tableau corresponds to an l - stable scheme ( that is , @xmath277 , being @xmath85 a vector whose components are all equal to @xmath119 ) , whereas the explicit tableau is ssp@xmath278 , where @xmath278 denotes the order of the ssp scheme . we shall use the notation ssp@xmath279 , where the triplet @xmath280 characterizes the number of @xmath281 stages of the implicit scheme , the number @xmath282 of stages of the explicit scheme and the order @xmath44 of the imex scheme .
there are different imex rk schemes available in the literature .
we have considered only third order imex schemes , some of them found in the literature @xcite and others developed by us .
all of them are based on a third order ssp explicit scheme that can be implemented efficiently by using only two levels of fields and one of rhs .
it is worth mentioning that these methods are still under development and have few drawbacks .
probably the most serious one is an accuracy degradation for some range of the relaxation time @xmath43 .
c c c c c c 0 & & 0 & 0 & 0 & 0 + 0 & & 0 & 0 & 0 & 0 + 1 & & 0 & 1 & 0 & 0 + 1/2 & & 0 & 1/4 & 1/4 & 0 + & & 0 & 1/6 & 1/6 & 2/3 + c c c c c c @xmath8 & & @xmath8 & 0 & 0 & 0 + 0 & & -@xmath8 & @xmath8 & 0 & 0 + 1 & & 0 & @xmath283 & @xmath8 & 0 + 1/2 & & @xmath284 & @xmath285 & @xmath286 & @xmath8 + & & 0 & 1/6 & 1/6 & 2/3 + @xmath287 [ ssp3 - 433 ] c c c c c c c 0 & & 0 & 0 & 0 & 0 & 0 + 0 & & 0 & 0 & 0 & 0 & 0 + 1 & & 0 & 1 & 0 & 0 & 0 + 1/2 & & 0 & 1/4 & 1/4 & 0 & 0 + 1 & & 0 & 1/6 & 1/6 & 2/3 & 0 + & & 0 & 1/6 & 1/6 & 2/3 & 0 + 1.0 cm c c c c c c c @xmath8 & & @xmath8 & 0 & 0 & 0 & 0 + 0 & & -@xmath8 & @xmath8 & 0 & 0 & 0 + 1 & & 0 & @xmath283 & @xmath8 & 0 & 0 + 1/2 & & @xmath288 & @xmath289 & @xmath290 & @xmath8 & 0 + 1 & & 0 & 1/6 & 0 & 2/3 & 1/6 + & & 0 & 1/6 & 0 & 2/3 & 1/6 + @xmath291 [ ssp3 - 533 ]
the ideal mhd limit can be obtained by requiring the current to be finite even in the limit of infinite isotropic conductivity , leading to the condition @xmath292 .
the ohm s law current becomes undetermined ( i.e. , an infinite conductivity multiplying a vanishing electric field in the co - moving frame ) , but it can still be computed from the redundant maxwell equation for the electric field evolution ( [ maxwellext_3 + 1_eq1a ] ) . the evolution of the magnetic field can be simplified by substituting the ideal mhd condition in ( [ maxwellext_3 + 1_eq1c ] ) , @xmath293 \nonumber \\ & = & \sqrt{\gamma } [ -b^k \partial_k \beta^i + \phi \gamma^{ik } ( \partial_k \alpha + \gamma^j_{jk})]\end{aligned}\ ] ] assuming an ideal gas eos , and after some manipulations in the definition of @xmath298 ( [ tmunu_decomposition2a ] ) , the resulting final equation to solve is @xmath299 x + [ \frac{(\gamma-1 ) \sqrt{c}}{\gamma } - 1 ] d \nonumber \\ & + & [ 1-\frac{c}{2 } ] b^2 - \frac{1}{2 x^2 } ( s_i b^i)^2 - \tau ~~.\end{aligned}\ ] ]
the author acknowledges his long time collaborators e .hirschmann , s. liebling and c .thompson for useful comments , and particularly to d. alic for discussions on the matching of the currents , d. neilsen for his help on implementing the imex in had , and l. lehner for carefully reading and discussing this manuscript .
this work was supported by the jeffrey l. bishop fellowship .
computations were performed in scinet . | this work presents an implementation of the _ resistive mhd _ equations for a generic algebraic ohm s law which includes the effects of finite resistivity within full general relativity .
the implementation naturally accounts for magnetic - field - induced anisotropies and , by adopting a phenomenological current , is able to accurately describe electromagnetic fields in the star and in its magnetosphere .
we illustrate the application of this approach in interesting systems with astrophysical implications ; the aligned rotator solution and the collapse of a magnetized rotating neutron star to a black hole .
+ [ firstpage ] mhd plasmas gravitation methods : numerical |
the method of _ stable random projections_@xcite , as an efficient tool for computing pairwise distances in massive high - dimensional data , provides a promising mechanism to tackle some of the challenges in modern machine learning . in this paper
, we provide an easy - to - implement algorithm for _ stable random projections _ which is both statistically accurate and computationally efficient .
we denote a data matrix by @xmath3 , i.e. , @xmath4 data points in @xmath5 dimensions .
data sets in modern applications exhibit important characteristics which impose tremendous challenges in machine learning @xcite : * modern data sets with @xmath6 or even @xmath7 points are not uncommon in supervised learning , e.g. , in image / text classification , ranking algorithms for search engines , etc . in the unsupervised domain ( e.g. , web clustering , ads clickthroughs ,
word / term associations ) , @xmath4 can be even much larger .
* modern data sets are often of ultra high - dimensions ( @xmath5 ) , sometimes in the order of millions ( or even higher ) , e.g. , image , text , genome ( e.g. , snp ) , etc .
for example , in image analysis , @xmath5 may be @xmath8 if using pixels as features , or @xmath9 million if using color histograms as features .
* modern data sets are sometimes collected in a dynamic streaming fashion . *
large - scale data are often heavy - tailed , e.g. , image and text data .
some large - scale data are dense , such as image and genome data . even for data sets which are sparse , such as text
, the absolute number of non - zeros may be still large .
for example , if one queries machine learning " ( a not - too - common term ) in google.com , the total number of pagehits is about 3 million . in other words ,
if one builds a term - doc matrix at web scale , although the matrix is sparse , most rows will contain large numbers ( e.g. , millions ) of non - zero entries .
many learning algorithms require a similarity matrix computed from pairwise distances of the data matrix @xmath3 .
examples include clustering , nearest neighbors , multidimensional scaling , and kernel svm ( support vector machines ) .
the similarity matrix requires @xmath10 storage space and @xmath11 computing time .
this study focuses on the @xmath0 distance ( @xmath12 ) .
consider two vectors @xmath13 , @xmath14 ( e.g. , the leading two rows in @xmath15 ) , the @xmath0 distance between @xmath13 and @xmath16 is @xmath17 note that , strictly speaking , the @xmath0 distance should be defined as @xmath18 .
because the power operation @xmath19 is the same for all pairs , it often makes no difference whether we use @xmath18 or just @xmath20 ; and hence we focus on @xmath20 . the radial basis kernel ( e.g. , for svm )
is constructed from @xmath20 @xcite : @xmath21 when @xmath22 , this is the gaussian radial basis kernel . here @xmath23 can be viewed as a _ tuning _ parameter .
for example , in their histogram - based image classification project using svm , @xcite reported that @xmath24 and @xmath25 achieved good performance .
for heavy - tailed data , tuning @xmath23 has the similar effect as term - weighting the original data , often a critical step in a lot of applications @xcite . for popular kernel svm solvers including the _ sequential minimal optimization ( smo _ ) algorithm@xcite ,
storing and computing kernels is the major bottleneck .
three computational challenges were summarized in @xcite : * * _ computing kernels is expensive _ * * * _ computing full kernel matrix is wasteful _ * efficient svm solvers often do not need to evaluate all pairwise kernels . * * _ kernel matrix does not fit in memory _ * storing the kernel matrix at the memory cost @xmath10 is challenging when @xmath26 , and is not realistic for @xmath27 , because @xmath28 consumes at least @xmath29 gbs memory .
a popular strategy in large - scale learning is to evaluate distances * * on the fly**@xcite .
that is , instead of loading the similarity matrix in memory at the cost of @xmath10 , one can load the original data matrix at the cost of @xmath30 and recompute pairwise distances on - demand .
this strategy is apparently problematic when @xmath5 is not too small . for high - dimensional data , either loading the data matrix in memory is unrealistic or computing distances on - demand becomes too expensive .
those challenges are not unique to kernel svm ; they are general issues in distanced - based learning algorithms .
the method of _ stable random projections _ provides a promising scheme by reducing the dimension @xmath5 to a small @xmath31 ( e.g. , @xmath32 ) , to facilitate compact data storage and efficient distance computations .
the basic procedure of _ stable random projections _ is to multiply @xmath3 by a random matrix @xmath33 ( @xmath34 ) , which is generated by sampling each entry @xmath35 i.i.d . from a symmetric stable distribution @xmath36 .
the resultant matrix @xmath37 is much smaller than @xmath15 and hence it may fit in memory .
suppose a stable random variable @xmath38 , where @xmath39 is the scale parameter .
then its characteristic function ( fourier transform of the density function ) is @xmath40 which does not have a closed - form inverse except for @xmath41 ( normal ) or @xmath42 ( cauchy ) .
note that when @xmath41 , @xmath39 corresponds to `` @xmath43 '' ( not `` @xmath44 '' ) in a normal .
corresponding to the leading two rows in @xmath15 , @xmath13 , @xmath14 , the leading two rows in @xmath45 are @xmath46 , @xmath47 .
the entries of the difference , @xmath48 for @xmath49 to @xmath31 , are i.i.d .
samples from a stable distribution with the scale parameter being the @xmath0 distance @xmath20 , due to properties of fourier transforms .
for example , when @xmath41 , a weighted sum of i.i.d .
standard normals is also normal with the scale parameter ( i.e. , variance ) being the sum of squares of all weights .
once we obtain the stable samples , one can discard the original matrix @xmath15 and the remaining task is to estimate the scale parameter @xmath20 for each pair . + some applications of _ stable random projections _
are summarized as follows : * * _ computing all pairwise distances _ * the cost of computing all pairwise distances of @xmath3 , @xmath11 , is significantly reduced to @xmath50 . * * _ estimating @xmath0 distances online _
* for @xmath26 , it is challenging or unrealistic to materialize all pairwise distances in @xmath15 .
thus , in applications such as online learning , databases , search engines , and online recommendation systems , it is often more efficient if we store @xmath51 in the memory and estimate any distance _ on the fly _ if needed . estimating distances online is the standard strategy in large - scale kernel learning@xcite . with _ stable random projections _ , this simple strategy becomes effective in high - dimensional data .
* * _ learning with dynamic streaming data _ * in reality , the data matrix may be updated overtime .
in fact , with streaming data arriving at high - rate@xcite , the `` data matrix '' may be never stored and hence all operations ( such as clustering and classification ) must be conducted on the fly .
the method of _ stable random projections _ provides a scheme to compute and update distances on the fly in one - pass of the data ; see relevant papers ( e.g. , @xcite ) for more details on this important and fast - developing subject . * * _ estimating entropy _ * the entropy distance @xmath52 is a useful statistic .
a workshop in nips03 ( www.menem.com/~ilya/pages/nips03 ) focused on entropy estimation .
a recent practical algorithm is simply using the difference between the @xmath53 and @xmath54 distances@xcite , where @xmath55 , @xmath56 , and the distances were estimated by _
stable random projections_. if one tunes the @xmath0 distances for many different @xmath23 ( e.g. , @xcite ) , then _
stable random projections _ will be even more desirable as a cost - saving device .
recall that the method of _ stable random projections _ boils down to a statistical estimation problem . that is , estimating the scale parameter @xmath20 from @xmath31 i.i.d .
samples @xmath57 , @xmath49 to @xmath31 .
we consider that a good estimator @xmath58 should have the following desirable properties : * ( asymptotically ) unbiased and small variance . * computationally efficient . *
exponential decrease of error ( tail ) probabilities .
the _ arithmetic mean _
estimator @xmath59 is good for @xmath60 . when @xmath61 , the task is less straightforward because ( 1 ) no explicit density of @xmath62 exists unless @xmath42 or @xmath63 ; and ( 2 ) @xmath64 only when @xmath65 . initially reported in arxiv in 2006 , @xcite proposed the _ geometric mean _ estimator where @xmath66 is the gamma function , and the _ harmonic mean _
estimator more recently , @xcite proposed the _ fractional power _
estimator where all three estimators are unbiased or asymptotically ( as @xmath67 ) unbiased .
figure [ fig_efficiency ] compares their asymptotic variances in terms of the cramr - rao efficiency , which is the ratio of the smallest possible asymptotic variance over the asymptotic variance of the estimator , as @xmath67 .
the _ geometric mean _ estimator , @xmath68 exhibits tail bounds in exponential forms , i.e. , the errors decrease exponentially fast : @xmath69 the _ harmonic mean _ estimator , @xmath70 , works well for small @xmath23 , and has exponential tail bounds for @xmath71 . the _ fractional power _
estimator , @xmath72 , has smaller asymptotic variance than both the _ geometric mean _ and _ harmonic mean _ estimators .
however , it does not have exponential tail bounds , due to the restriction @xmath73 in its definition . as shown in @xcite
, it only has finite moments slightly higher than the @xmath74 order , when @xmath23 approaches 2 ( because @xmath75 ) , meaning that large errors may have a good chance to occur .
we will demonstrate this by simulations . in the definitions of @xmath68 , @xmath70 and @xmath72 , all three estimators require evaluating fractional powers , e.g. , @xmath76 .
this operation is relatively expensive , especially if we need to conduct this tens of billions of times ( e.g. , @xmath77 ) .
for example , @xcite reported that , although the radial basis kernel ( [ eqn_rbs_kernel ] ) with @xmath25 achieved good performance , it was not preferred because evaluating the square root was too expensive .
we propose the _ optimal quantile _ estimator , using the @xmath78th smallest @xmath79 : @xmath80 where @xmath81 is chosen to minimize the asymptotic variance .
this estimator is computationally attractive because * selecting * should be much less expensive than evaluating fractional powers .
if we are interested in @xmath82 instead , then we do not even need to evaluate any fractional powers .
as mentioned , in many cases using either @xmath20 or @xmath82 makes no difference and @xmath20 is often preferred because it avoids taking @xmath19 power
. the radial basis kernel ( [ eqn_rbs_kernel ] ) requires @xmath20 .
thus this study focuses on @xmath20 . on the other hand ,
if we can estimate @xmath82 directly , for example , using ( [ eqn_d_oq ] ) without the @xmath23th power , we might as well just use @xmath82 if permitted . in case
we do not need to evaluate any fractional power , our estimator will be even more computationally efficient .
in addition to the computational advantages , this estimator also has good theoretical properties , in terms of both the variances and tail probabilities : 1 .
figure [ fig_efficiency ] illustrates that , compared with the _ geometric mean _ estimator , its asymptotic variance is about the same when @xmath83 , and is considerably smaller when @xmath2 . compared with the _ fractional power _
estimator , it has smaller asymptotic variance when @xmath84 .
in fact , as will be shown by simulations , when the sample size @xmath31 is not too large , its mean square errors are considerably smaller than the _ fractional power _
estimator when @xmath85 .
the _ optimal quantile _
estimator exhibits tail bounds in exponential forms .
this theoretical contribution is practically important , for selecting the sample size @xmath31 . in learning theory ,
the generalization bounds are often loose . in our case , however , the bounds are tight because the distribution is specified
. the next section will be devoted to analyzing the _ optimal quantile _ estimator .
recall the goal is to estimate @xmath20 from @xmath86 , where @xmath87 , i.i.d .
since the distribution belongs to the scale family , one can estimate the scale parameter from quantiles . due to symmetry , it is natural to consider the absolute values : @xmath88 which is best understood by the fact that if @xmath89 , then @xmath90 , or more obviously , if @xmath91 , then @xmath92 . by properties of order statistics @xcite , any @xmath93-quantile will provide an asymptotically unbiased estimator .
lemma [ lem_var_q ] provides the asymptotic variance of @xmath94 .
[ lem_var_q ] denote @xmath95 and @xmath96 the probability density function and the cumulative density function of @xmath97 , respectively .
the asymptotic variance of @xmath94 defined in ( [ eqn_quantile ] ) is @xmath98 where @xmath99 .
* proof : * see appendix [ proof_lem_var_q ] . @xmath100 .
we choose @xmath102 so that the asymptotic variance ( [ eqn_var_q ] ) is minimized , i.e. , @xmath103 the convexity of @xmath104 is important .
graphically , @xmath104 is a convex function of @xmath93 , i.e. , a unique minimum exists . an algebraic proof , however , is difficult .
nevertheless , we can obtain analytical solutions when @xmath42 and @xmath105 .
[ lem_convexity ] when @xmath42 or @xmath105 , the function @xmath104 defined in ( [ eqn_g ] ) is a convex function of @xmath93 .
when @xmath42 , the optimal @xmath106 .
when @xmath107 , @xmath108 is the solution to @xmath109 .
* proof : * see appendix [ proof_lem_convexity ] . @xmath100 .
it is also easy to show that when @xmath41 , @xmath110 .
we denote the _ optimal quantile _ estimator by @xmath111 , which is same as @xmath112 . for general @xmath23
, we resort to numerical solutions , as presented in figure [ fig_opt_quantile ] .
although @xmath111 ( i.e. , @xmath112 ) is asymptotically ( as @xmath67 ) unbiased , it is seriously biased for small @xmath31 .
thus , it is practically important to remove the bias . the unbiased version of the _ optimal quantile _ estimator is @xmath113 where @xmath114 is the expectation of @xmath111 at @xmath115 . for @xmath42 , @xmath63 , or @xmath116 , we can evaluate the expectations ( i.e. , integrals ) analytically or by numerical integrations . for general @xmath23 , as the probability density is not available , the task is difficult and prone to numerical instability . on the other hand ,
since the monte - carlo simulation is a popular alternative for evaluating difficult integrals , a practical solution is to simulate the expectations , as presented in figure [ fig_bias ] .
figure [ fig_bias ] illustrates that @xmath117 , meaning that this correction also reduces variance while removing bias ( because @xmath118 ) .
for example , when @xmath119 and @xmath120 , @xmath121 , which is significant , because @xmath122 implies a @xmath123 difference in terms of variance , and even more considerable in terms of the mean square errors mse = variance + bias@xmath124 .
@xmath114 can be tabulated for small @xmath31 , and absorbed into other coefficients , i.e. , this does not increase the computational cost at run time .
we fix @xmath114 as reported in figure [ fig_bias ] . the simulations in section [ sec_simulations ]
directly used those fixed @xmath114 values .
figure [ fig_compu_ratio ] compares the computational costs of the _ geometric mean _ , the _ fractional power _ , and the _ optimal quantile _ estimators .
the _ harmonic mean _
estimator was not included as it costs very similarly to the _ fractional power _ estimator .
we used the build - in function `` pow''in gcc for evaluating the fractional powers .
we implemented a `` quick select '' algorithm , which is similar to quick sort and requires on average linear time . for simplicity ,
our implementation used recursions and the middle element as pivot . also , to ensure fairness , for all estimators , coefficients which are functions of @xmath23 and/or @xmath31 were pre - computed .
normalized by the computing time of @xmath68 , we observe that relative computational efficiency does not strongly depend on @xmath23 .
we do observe that the ratio of computing time of @xmath68 over that of @xmath125 increases consistently with increasing @xmath31 .
this is because in the definition of @xmath111 ( and hence also @xmath125 ) , it is required to evaluate the fractional power once , which contributes to the total computing time more significantly at smaller @xmath31 .
figure [ fig_compu_ratio ] illustrates that , ( a ) the _ geometric mean _ estimator and the _ fractional power _
estimator are similar in terms of computational efficiency ; ( b ) the _ optimal quantile _ estimator is nearly one order of magnitude more computationally efficient than the _ geometric mean _ and _ fractional power _ estimators .
because we implemented a `` nave '' version of `` quick select '' using recursions and simple pivoting , the actual improvement may be more significant .
also , if applications require only @xmath82 , then no fractional power operations are needed for @xmath125 and the improvement will be even more considerable .
error ( tail ) bounds are essential for determining @xmath31 .
the variance alone is not sufficient for that purpose .
if an estimator of @xmath39 , say @xmath126 , is normally distributed , @xmath127 , the variance suffices for choosing @xmath31 because its error ( tail ) probability @xmath128 is determined by @xmath129 . in general
, a reasonable estimator will be asymptotically normal , for small enough @xmath130 and large enough @xmath31 . for a finite @xmath31 and a fixed @xmath130 , however , the normal approximation may be ( very ) poor .
this is especially true for the _ fractional power _
estimator , @xmath72 .
thus , for a good motivation , lemma [ lem_bounds ] provides the error ( tail ) probability bounds of @xmath94 for any @xmath93 , not just the optimal quantile @xmath78 .
[ lem_bounds ] denote @xmath131 and its probability density function by @xmath132 and cumulative function by @xmath133 .
given @xmath87 , i.i.d .
, @xmath49 to @xmath31 . using @xmath94 in ( [ eqn_quantile ] ) , then @xmath134 @xmath135 as @xmath136 * proof : * see appendix [ proof_lem_bounds ] . @xmath100 the limit in ( [ eqn_g_rl_limit ] ) as @xmath137 is precisely twice the asymptotic variance factor of @xmath94 in ( [ eqn_var_q ] ) , consistent with the normality approximation mentioned previously .
this explains why we express the constants as @xmath138 .
( [ eqn_g_rl_limit ] ) also indicates that the tail bounds achieve the `` optimal rate '' for this estimator , in the language of large deviation theory . by the bonferroni bound ,
it is easy to determine the sample size @xmath31 @xmath139 [ lem_jl_quantile ] using @xmath94 with @xmath140 , any pairwise @xmath0 distance among @xmath4 points can be approximated within a @xmath141 factor with probability @xmath142 .
it suffices to let @xmath143 , where @xmath144 , @xmath145 are defined in lemma [ lem_bounds ] .
the bonferroni bound can be unnecessarily conservative .
it is often reasonable to replace @xmath146 by @xmath147 , meaning that except for a @xmath148 fraction of pairs , any distance can be approximated within a @xmath141 factor with probability @xmath149 .
figure [ fig_bounds ] plots the error bound constants for @xmath150 , for both the recommended _ optimal quantile _
estimator @xmath111 and the baseline _ sample median _
estimator @xmath151 .
although we choose @xmath111 based on the asymptotic variance , it turns out @xmath111 also exhibits ( much ) better tail behaviors ( i.e. , smaller constants ) than @xmath151 , at least in the range of @xmath150 . consider @xmath152 ( recall we suggest replacing @xmath153 by @xmath154 ) , with @xmath155 , @xmath156 , and @xmath157 .
because @xmath158 around @xmath159 , we obtain @xmath160 , which is still a relatively large number ( although the original dimension @xmath5 might be @xmath161 ) .
if we choose @xmath162 , then approximately @xmath163 .
it is possible @xmath164 might be still conservative , for three reasons : ( a ) the tail bounds , although sharp , " are still upper bounds ; ( b ) using @xmath165 is conservative because @xmath166 is usually much smaller than @xmath167 ; ( c ) this type of tail bounds is based on relative error , which may be stringent for small ( @xmath168 ) distances .
in fact , some earlier studies on _ normal random projections _ ( i.e. , @xmath60 ) @xcite empirically demonstrated that @xmath169 appeared sufficient .
we resort to simulations for comparing the finite sample variances of various estimators and assessing the more precise error ( tail ) probabilities .
one advantage of _ stable random projections _ is that we know the ( manually generated ) distributions and the only source of errors is from the random number generations .
thus , we can simply rely on simulations to evaluate the estimators without using real data .
in fact , after projections , the projected data follow exactly the stable distribution , regardless of the original real data distribution . without loss of generality ,
we simulate samples from @xmath170 and estimate the scale parameter ( i.e. , 1 ) from the samples . repeating the procedure @xmath171 times , we can reliably evaluate the mean square errors ( mse ) and tail probabilities . as illustrated in figure [ fig_simu_mse ] , in terms of the mse , the _ optimal quantile _
estimator @xmath125 outperforms both the _ geometric mean _ and _ fractional power _
estimators when @xmath85 and @xmath172 .
the _ fractional power _ estimator does not appear to be very suitable for @xmath85 , especially for @xmath23 close to 2 , even when the sample size @xmath31 is not too small ( e.g. , @xmath32 ) . for @xmath83 , however , the _ fractional power _
estimator has good performance in terms of mse , even for small @xmath31 .
figure [ fig_simu_tail ] presents the simulated right tail probabilities , @xmath173 , illustrating that when @xmath2 , the _ fractional power _ estimator can exhibit very bad tail behaviors . for @xmath174 ,
the _ fractional power _ estimator demonstrates good performance at least for the probability range in the simulations .
thus , figure [ fig_simu_tail ] demonstrates that the _ optimal quantile _ estimator consistently outperforms the _ fractional power _ and the _ geometric mean _
estimators when @xmath85 .
there have been many studies of _ normal random projections _ in machine learning , for dimension reduction in the @xmath175 norm , e.g. , @xcite , highlighted by the johnson - lindenstrauss ( jl ) lemma @xcite , which says @xmath176 suffices when using normal ( or normal - like , e.g. , @xcite ) projection methods .
the method of _ stable random projections _ is applicable for computing the @xmath0 distances ( @xmath1 ) , not just for @xmath175 .
* lemma 1 , lemma 2 , theorem 3 ) suggested the _ median _
( i.e. , @xmath177 quantile ) estimator for @xmath178 and argued that the sample complexity bound should be @xmath179 ( @xmath180 in their study ) .
their bound was not provided in an explicit form and required an `` @xmath130 is small enough '' argument . for @xmath181 , (
* lemma 4 ) only provided a conceptual algorithm , which `` is not uniform . '' in this study , we prove the bounds for any @xmath93-quantile and any @xmath12 ( not just @xmath182 ) , in explicit exponential forms , with no unknown constants and no restriction that `` @xmath130 is small enough . ''
the quantile estimator for stable distributions was proposed in statistics quite some time ago , e.g. , @xcite .
@xcite mainly focused on @xmath183 and recommended using @xmath184 quantiles ( mainly for the sake of smaller bias ) .
@xcite focused on @xmath185 and recommended @xmath186 quantiles .
this study considers all @xmath1 and recommends @xmath93 based on the minimum asymptotic variance . because the bias can be easily removed ( at least in the practical sense ) , it appears not necessary to use other quantiles only for the sake of smaller bias .
tail bounds , which are useful for choosing @xmath93 and @xmath31 based on confidence intervals , were not available in @xcite .
finally , one might ask if there might be better estimators . for @xmath42 , @xcite proposed using a linear combination of quantiles ( with carefully chosen coefficients ) to obtain an asymptotically optimal estimator for the cauchy scale parameter . while it is possible to extend their result to general @xmath187 ( requiring some non - trivial work ) , whether or not it will be practically better than the _ optimal quantile _ estimator is unclear because the extreme quantiles severely affect the tail probabilities and finite - sample variances and hence some kind of truncation ( i.e. , discarding some samples at extreme quantiles ) is necessary . also , exponential tail bounds of the linear combination of quantiles may not exist or may not be feasible to derive .
in addition , the _ optimal quantile _ estimator is computationally more efficient .
many machine learning algorithms operate on the training data only through pairwise distances .
computing , storing , updating and retrieving the `` matrix '' of pairwise distances is challenging in applications involving massive , high - dimensional , and possibly streaming , data .
for example , the pairwise distance matrix can not fit in memory when the number of observations exceeds @xmath161 ( or even @xmath188 ) .
the method of _ stable random projections _ provides an efficient mechanism for computing pairwise distances using low memory , by transforming the original high - dimensional data into _ sketches _ , i.e. , a small number of samples from @xmath23-stable distributions , which are much easier to store and retrieve .
this method provides a uniform scheme for computing the @xmath0 pairwise distances for all @xmath12 . choosing an appropriate @xmath23 is often critical to the performance of learning algorithms . in principle
, we can tune algorithms for many @xmath0 distances ; and _ stable random projections _ can provide an efficient tool . to recover the original distances ,
we face an estimation task . compared with previous estimators based on the _ geometric mean _ , _ the harmonic mean _ , or the _ fractional power _
, the proposed _ optimal quantile _
estimator exhibits two advantages .
firstly , the _ optimal quantile _ estimator is nearly one order of magnitude more efficient than other estimators ( e.g. , reducing the training time from one week to one day ) .
secondly , the _ optimal quantile _ estimator is considerably more accurate when @xmath85 , in terms of both the variances and error ( tail ) probabilities .
note that @xmath189 corresponds to a convex norm ( satisfying the triangle inequality ) , which might be another motivation for using @xmath0 distances with @xmath189 .
one theoretical contribution is the explicit tail bounds for general quantile estimators and consequently the sample complexity bound @xmath176 .
those bounds may guide practitioners in choosing @xmath31 , the number of projections .
the ( practically useful ) bounds are expressed in terms of the probability functions and hence they might be not as convenient for further theoretical analysis . also , we should mention that the bounds do not recover the optimal bound of the _ arithmetic mean _ estimator when @xmath41 , because the _ arithmetic mean _ estimator is statistically optimal at @xmath60 but the _ optimal quantile _ estimator is not . while we believe that applying _
stable random projections _ in machine learning has become straightforward , there are interesting theoretical issues for future research .
for example , how theoretical properties of learning algorithms may be affected if the approximated ( instead of exact ) @xmath0 distances are used ?
denote @xmath95 and @xmath96 the probability density function and the cumulative density function of @xmath97 , respectively .
similarly we use @xmath190 and @xmath191 for @xmath192 . due to symmetry , the following relations hold @xmath193 let @xmath194 and @xmath195 .
then , following known statistical results , e.g. , ( * ? ? ?
* theorem 9.2 ) , the asymptotic variance of @xmath196 should be @xmath197 by delta method , "
i.e. , @xmath198 , @xmath199 first , consider @xmath42 . in this case , @xmath200 it suffices to study @xmath201 .
@xmath202 because @xmath203 for @xmath204 , it is easy to see that @xmath205 , and @xmath206 . thus , @xmath207 , i.e. , @xmath208 is convex and so is @xmath209 .
since @xmath210 , we know @xmath106 .
+ next we consider @xmath105 , using a fact @xcite that as @xmath211 , @xmath212 converges to @xmath213 , where @xmath214 stands for an exponential distribution with mean 1 .
denote @xmath215 and @xmath216 .
the sample quantile estimator becomes @xmath217 in this case , it is straightforward to show that @xmath218 is a convex function of @xmath93 and the minimum is attained by solving @xmath219 , i.e. , @xmath220 . given @xmath31 i.i.d .
samples , @xmath87 , @xmath49 to @xmath31 .
let @xmath221 , @xmath49 to @xmath31 .
denote by @xmath222 the cumulative density of @xmath223 , and by @xmath224 the empirical cumulative density of @xmath223 , @xmath49 to @xmath31 .
it is the basic fact@xcite about order statistics that @xmath225 follows a binomial , i.e. , @xmath226 . for simplicity ,
we replace @xmath227 by @xmath228 , @xmath229 by @xmath230 , and @xmath20 by @xmath39 , in this proof .
consider the general quantile estimator @xmath94 defined in ( [ eqn_quantile ] ) .
for @xmath235 , ( again , denote @xmath236 ) , @xmath237 where @xmath238 and @xmath239 .
thus @xmath240^k%\\\notag & % = & \exp\left ( % -k\left(-(1-q)\log\left(1-f\left ( \left((1+\epsilon)\right)^{1/\alpha}w;1\right)\right ) - q \log \left(f\left ( \left((1+\epsilon)\right)^{1/\alpha}w;1\right)\right ) + ( 1-q)\log ( 1-q ) + q\log(q)\right)\right)\\\notag = \exp\left(-k\frac{\epsilon^2}{g_{r , q}}\right).\end{aligned}\ ] ] where for @xmath241 , @xmath242 where @xmath243 and @xmath244 .
thus , @xmath245^k%\\\notag & = \exp\left(-k\frac{\epsilon^2}{g_{l , q}}\right),\end{aligned}\ ] ] where chernoff , h. , gastwirth , j.l . , johns , m.v .
: asymptotic distribution of linear combinations of functions of order statistics with applications to estimation .
the annals of mathematical statistics * 38*(1 ) ( 1967 ) 5272 | the method of _ stable random projections _ is a tool for efficiently computing the @xmath0 distances using low memory , where @xmath1 is a tuning parameter .
the method boils down to a statistical estimation task and various estimators have been proposed , based on the _ geometric mean _ , the _ harmonic mean _ , and the _ fractional power _ etc .
this study proposes the * _ optimal quantile _ * estimator , whose main operation is * _ selecting _ * , which is considerably less expensive than taking fractional power , the main operation in previous estimators .
our experiments report that the _ optimal quantile _ estimator is nearly one order of magnitude more computationally efficient than previous estimators . for large - scale learning tasks
in which storing and computing pairwise distances is a serious bottleneck , this estimator should be desirable .
in addition to its computational advantages , the _ optimal quantile _ estimator exhibits nice theoretical properties .
it is more accurate than previous estimators when @xmath2 .
we derive its theoretical error bounds and establish the explicit ( i.e. , no hidden constants ) sample complexity bound . |
consider a two state particle , _
e.g. _ a two level atom , moving along a one dimensional lattice . the displacement is conditional ; that is , the particle jumps one site at each time to the right or to the left depending on its internal state .
the movement of such a particle is rather trivial : if the particle is in its upor down internal state it moves to the right or to the left , respectively , and it undergoes a superposition of right and left displacements if it is in a superposition of its up and down states .
but the movement becomes nontrivial if after each displacement a suitable unitary transformation , _
e.g. _ the hadamard transformation , leads to a superposition of the internal states of the particle .
now the probability distribution that the particle would be found at a given position is quite rich : it is null at odd ( even ) positions at odd ( even ) times , shows two strong peaks near the right and left edges , develops strong oscillations near these peaks and has a smooth plateau in the central region .
this process is known as the ( coined ) quantum walk . in a sense ,
the quantum walk , which we refer to as qw , is the quantum generalization of the classical random walk , and as such was first introduced ten years ago by aharonov _
it was later rediscovered by meyer @xcite in the context of quantum cellular automata , and more recently by watrous @xcite in the context of quantum algorithms .
the qw introduced in these papers is called the _
discrete _ qw ( be it coined @xcite or not @xcite ) , and must be distinguished from the so called _ continuous _ qw , introduced by farhi and gutmann in 1998 @xcite , which realizes to some extent a quantum version of continuous time classical markov chains .
the connection between these two types of qws is still unclear @xcite .
the qw , both in its discrete and continuous versions , is a process that has recently attracted the attention of many workers in the quantum computing community ; see , for example , the recent review by julia kempe [ 5 ] .
of interest is the possibility that qws will give rise to new quantum algorithms that show clear speedups over classical algorithms .
the first results in this direction have been obtained by shenvi _
shenvi , who have shown that a qw can perform the same tasks as grover s search algorithm @xcite , and by childs _
@xcite , who have introduced an algorithm for crossing a special type of graph exponentially faster than can be done classically . in the present communication
we give an alternate description of the discrete qw that we feel gives physical insight into the nature of that process .
this paper is an extention of our previous work @xcite , in which we showed that the qw on the line is not a characteristically quantum process in the specific sense that it can be understood simply as a wave phenomenon , thus opening the possibility of entirely classical implementations @xcite .
indeed , wave implementations hillery , jeong of the qw have already been discussed . in other words ,
the characteristic quantum feature of entanglement does not seem to play any central role in the qw , at least in the unidimensional case .
the situation could be different for higher dimensional qws ; see the discussion in knight _ et al . _
@xcite for some insight into the role of entanglement in higher dimensional qws .
the rest of the article is organized as follows . in section 2
we give the evolution equations for the standard discrete coined qw , and reformulate them in a way that directly exhibits an important property : the two coin states evolve independently of each other , the coupling between them entering only through the first iteration of the walk . in section 3
we derive the formal solution of the qw in a manner different to those already carried out @xcite .
this alternative derivation allows a rigorous derivation of the long wavelength approximation that corresponds to the continuous limit of the qw we pointed out earlier @xcite . in section 4
we give the solution of this long wavelength limit and discuss some of its properties .
finally in section 5 we give our conclusions .
let @xmath0 be the hilbert space of the particle positions and @xmath1 a basis of @xmath2 ; and let @xmath3 be the hilbert space describing the two states of the qubit associated with the internal states of the particle ( usually referred to as the coin ) , and @xmath4 a basis of @xmath3 .
the state of the total system belongs to the space @xmath5 .
the dynamics of the system is governed by two operations , the conditional diplacement @xmath6and the transformation acting on the internal states of the coin @xmath7@xmath8 , which is the most general unitary transformation that one needs to consider @xcite ( notice that @xmath9 is the hadamard transformation when @xmath10 ) . the qw is implemented by the repeated action of the operator @xmath11 , _
i.e. _ if @xmath12 denotes the state of the system after @xmath13 iterations then @xmath14which can be written as @xmath15 , \]]with the equations of evolution of the probability amplitudes given by @xmath16finally , the probability of finding the particle at position @xmath17 at iteration ( time ) @xmath13 , is given by @xmath18 eqs.([amp1 ] ) and ( [ amp2 ] ) are the discrete qw equations as can be found _
e.g. _ in the work of carteret @xcite for the specific case of the fair coin @xmath10 .
different equations of evolution , but still leading to the same final probability , can be found by defining the walk as the repeated action of the operator @xmath19 , instead of @xmath20 , as is done by nayak and vishwanath @xcite . in the appendix we identify the transformations that connect these different conventions for the probability amplitudes .
there is a crucial feature in eqs.([amp1 ] ) and ( [ amp2 ] ) that is at the heart of our treatment of the qw . substituting eq.([amp2 ] ) into eq.([amp1 ] ) and rearranging the terms one easily gets @xmath21noticing that the last term in the right hand side is nothing but @xmath22 , and making the change @xmath23 , we can write @xmath24a similar equation for @xmath25 can be derived .
that is , eqs.([amp1 ] ) and ( [ amp2 ] ) are equivalent to @xmath26 , \;\;\;a = r , l \label{amp3}\]]which is a remarkable equation , as it shows that the evolutions of the two coin states are independent .
that is , @xmath27 does not depend on @xmath25 at previous times , and _ vice versa . _
the only coupling between @xmath28 and @xmath25 appears at the first interation .
notice that in eq .
( [ amp3 ] ) one needs to specify two initial conditions ( both @xmath29 and @xmath30 ) and one needs eqs.([amp1 ] ) and ( [ amp2 ] ) for calculating @xmath30 from @xmath29 .
but apart from this initial coupling , the evolution is completely independent for the two coin states . in other words : the qw is composed of two independent qws , one for each coin state , coupled only at initialization .
eq.([amp3 ] ) is interesting for another reason already discussed in knight in the special case of the hadamard transformation ( @xmath10 ) .
simple inspection leads to the idea that eq.([amp3 ] ) can be understood as the discretization of a differential equation involving all the odd derivatives with respect to space and time of a continuous field @xmath31@xmath32this is too nave an approximation , as the above equation does not respect the symetries of the solutions of eq.([amp3 ] ) : eq.([aprox ] ) describes the propagation of waves in the positive position direction , whilst the probability distribution of the quantum walk clearly shows propagation in both the positive and negative directions .
but it is clear that eq.([amp3 ] ) suggests the existence of some kind of continuous description of the discrete coined qw .
earlier @xcite we obtained such a continuous description by introducing , without justification , new amplitudes @xmath33 through the definition @xmath34 and then we derived continuous differential equations for the continuous fields @xmath35 .
these equations are similar to eq.(aprox ) but with a velocity that differs in its sign for the fields @xmath36 and @xmath37 . from these equations we obtained an approximate evolution equation by neglecting higher order derivatives @xmath38 a^{\pm } \left ( \xi , \tau \right ) , \label{aprox2}\]]with @xmath39 and @xmath40 .
eq.([aprox ] ) can be solved explicitly in terms of airy functions , and shows that to this degree of approximation the quantum walk fields @xmath33 satisfy the same equation as the evolution of an optical pulse in a medium with third order dispersion .
the weak points of this approach @xcite are that the fields @xmath41 are introduced without justification , that only the special case of the hadamard transformation is considered , and that the continuum limit is not clearly derived .
the goal of this paper is to present a rigorous derivation of eq.([aprox ] ) for arbitrary values of @xmath42 . for that purpose ,
we first turn to a new derivation of the full solution of the quantum walk on the line .
as at a given time identified by the index @xmath13 the amplitudes @xmath43 depend on the discrete variable @xmath17 .
so it is usual @xcite to define a continuous function of @xmath44 , periodic over a range of @xmath45 , such that the @xmath43 are the fourier components of that function @xmath46,@xmath47it is then possible to formulate the problem in terms of the continuous function @xmath48 of @xmath44 rather than the discrete function @xmath49 of @xmath17 .
a similar strategy can be implemented with respect to the dependence on the time step @xmath13 .
we essentially follow this approach below , but our perspective is slightly different . because we want to be able to understand eq .
( [ amp3 ] ) as the discretization of certain linear partial differential equations , we introduce fourier expansions such as ( [ fexpan ] ) in a more physical way . from the amplitudes @xmath43 defined at discrete points in space , labeled by @xmath17 , and time , labeled by @xmath13
, we can formally introduce a field @xmath50 at arbitrary points in space and time defined by @xmath51which produces appropriately weighted delta functions at the positions in space and time where@xmath43 is defined . here
@xmath52 is the distance between the sites in the one dimensional lattice , and @xmath53 is the time between steps in the walk .
we can formally introduce the fourier transform of this function @xmath54where from eq.([aux0 ] ) we find that @xmath55next , note that we can write each @xmath56 and @xmath57 uniquely as @xmath58 with @xmath59 and @xmath60 integers , and @xmath61 and @xmath62 lying within the ranges @xmath63 then using the identity @xmath64 which follows immediately from identifying the fourier sum representing the periodic function of @xmath65 on the left - hand side , we easily find from eq.(aux1 ) that @xmath66 .
then , making use of eqs.([aux0])([aux1 ] ) and ( [ didentity ] ) , we can write @xmath67 where @xmath68 _ cf . _ eq.([aux1 ] ) .
returning now to the discrete qw eqs.([amp3 ] ) , we multiply each side by @xmath69 @xmath70 and sum over @xmath17 and @xmath71 this yields the equation @xmath72which identifies the dispersion relations @xmath73 . because of the range of @xmath61 and @xmath62 appearing in eq.([abardef ] ) , we need only consider such solutions over the ranges ( [ range ] ) .
there are then two such solutions .
one solution , which we label @xmath74 , is that for which @xmath75 as @xmath76 .
the second , which we label @xmath77 , is given by @xmath78for @xmath61 respectively @xmath79 and @xmath80 .
these relations are represented in fig.1 .
solutions of eqs.([amp3 ] ) must respect the dispersion relations , and thus for eq.([a(k , omega ) ] ) to represent an actual solution we must have @xmath81 , \]]for some amplitudes @xmath82 and @xmath83 .
thus from eq.([abardef ] ) we find @xmath84using this expression for @xmath85 in eq.([a(x , t ) ] ) for @xmath50 , employing eq.([disp2 ] ) to write @xmath77 in terms of @xmath86 , and recalling that @xmath87 we can write @xmath88where @xmath89obviously , we can also write eq.([abar ] ) in the form @xmath90which will be used in the next section .
clearly the fields @xmath91 and @xmath92 represent disturbances propagating ( at least for small @xmath61 ) to the right and the left respectively , both with a dispersion relation @xmath74 .
the effect of the second dispersion relation @xmath77 is to introduce a kind of `` temporally antiferromagnetic '' behavior , indicated by the @xmath93 , associated with the leftward going wave .
the solution seems the simplest when formulated in terms of these two fields @xmath91 and @xmath92 .
since we have dynamical equations , eqs.([amp3 ] ) , for both the @xmath27 and the @xmath94 , we will have four fields in all as @xmath95 ranges over @xmath96 and @xmath97 : @xmath98 , @xmath99 , @xmath100 , and @xmath101 .
eqs.([amp3 ] ) requires two sets of initial conditions : the value of @xmath28 and @xmath25 at both @xmath102 and @xmath103 .
we can now see how these are built into the fields @xmath104 .
we begin by noting that from eqs.(aux0 ) and ( [ axtsol ] ) we can immediately identify @xmath105so to construct the @xmath106 appropriate for our initial conditions we must invert eqs.([axtft ] ) to find the @xmath106 in terms of @xmath107 .
this is easily done by focusing on points @xmath108 ; writing eqs.(axtft ) at such points and performing the sums indicated below , we find @xmath109as the left hand side is independent of time , so must be the right hand side . writing down these equations for @xmath110 and @xmath111 we
can then use eqs.([am0 ] ) and ( [ am1 ] ) to solve for the @xmath106 in terms of the @xmath112 and the @xmath113 .
the result is @xmath114 } e^{-ikmx } \notag \\ & = & \frac{1}{4\pi } x\sum_{m=-\infty } ^{\infty } \frac{e^{\pm i\omega _ { 0}(k)t}a_{m,0}\pm a_{m,1}}{\sqrt{1-\rho \sin ^{2}kx}}e^{-ikmx } , \label{a(k)}\end{aligned}\]]where in the second expression we have used the dispersion relation ( disp ) . the solution of the walk problem subject to arbitary initial conditions can then be written in the form of eq.([axtsol ] ) or , identifying the discrete points of interest in space and time , in the form @xmath115where @xmath104 are given by eqs.([axtft ] ) with the fourier components given by eq.([a(k ) ] ) . inserting those fourier components we find that we can write the full solution for @xmath104 in terms of a single green function @xmath116 , @xmath117where @xmath118and we have used the fact that @xmath119 . using the original walk equations ( [ amp1 ] ) and ( [ amp2 ] ) we
can then find the @xmath113 in terms of the @xmath29 , for @xmath120 .
then our final solution is given by eq.([finres ] ) with @xmath121and @xmath122 earlier , nayak and vishwanath @xcite formally solved the discrete qw and analyzed the asymptotic behaviour of the solution .
in the appendix we show that our formal solution and theirs are the same .
more recently , carteret _ et al .
_ @xcite have reviewed previous results concerning the asymptotics of the qw and presented new asymptotics .
of course , all those asymptotic results fully apply to our solution ; we do not pursue this further here .
the solution in the form we have constructed leads naturally to a passage to the continuum limit , which we identify by focusing on the continuous functions of space and time @xmath85 , and constructing simplifed equations for their evolution in the long wavelength limit
. our starting points are eqs.([axtft ] ) and ( [ abar2 ] ) , which we write again here for clarity:@xmath123with @xmath124 given by eqs.([a(k ) ] ) .
we consider the long wavelength limit by neglecting high frequency ( and small wavelength ) effects .
for that purpose we introduce low frequency fields @xmath125which in the long wavelength limit take the place of the fields @xmath107 . here
@xmath126 is a function that vanishes at high spatial frequencies @xmath61 ; that is , it plays the role of a low - pass filter . the cut - off that it provides
is assumed to be such that we can make a series expansion in @xmath127 .
taking the dispersion relation eq.([disp ] ) and expanding as a power series one gets @xmath128using the lowest order , @xmath129 , we approximate @xmath130 and write @xmath131 with @xmath132now , we take this aproximate dispersion relation in eq.([axt ] ) and write @xmath133where @xmath127 has been approximated by @xmath134 and the integration limits have been extended to infinity . finally from eqs.([axt2 ] ) it is easy to derive the equations of evolution of @xmath135 .
the equations can be written in the form @xmath136 a^{\pm } \left ( \xi , \tau \right ) \label{diff.eq.}\]]where we have introduced the new functions @xmath137 , and the normalized time and space variables @xmath138 and @xmath139 . notice that the form of this equation is independent of the particular form chosen for the cut - off function @xmath140 .
this is the differential equation derived earlier ( eq.(15 ) of @xcite ) for the special case @xmath10 .
it can be identified with the equation the envelope function of a light pulse would satisfy in a description of propagation through a material medium with no group velocity dispersion at the lowest order
such lowest order dispersion would appear through a term involving a second derivative with respect to @xmath141 but with higher order dispersion described by the third derivative term .
the equation is well known in fiber optics , where it is used to describe pulse propagation near the zero - dispersion wavelength @xcite .
notice that the effective dispersion coefficient in eq .
( [ diff.eq . ] ) is proportional to @xmath142 and thus , interestingly , can be varied by modifying the unitary transformation applied in the qw after each displacement .
let us proceed now to analyze the solution of eq.([diff.eq . ] ) .
by fourier transforming eq.([diff.eq . ] ) one easily gets @xmath143with @xmath144 the initial conditions @xmath145 can be calculated from eqs.([a(k ) ] ) and ( [ axt2 ] ) .
first we rewrite eqs.([a(k ) ] ) in the long wavelength limit @xmath146where we have approximated @xmath147 \approx 1 $ ] . substituting this into eq.([axt2 ] ) , and using the normalized spatial frequency @xmath148 we get @xmath149now we need to choose a particular cut off function . to obtain an explicit solution of eq .
( [ diff.eq . ] ) we adopt a gaussian , @xmath150where @xmath151 is a free parameter .
we further consider the usual case that the qw starts with the particle at position @xmath152 , and we get @xmath153 , \notag\end{aligned}\]]with @xmath154 a normalization factor that can be fixed by impossing @xmath155 = 1,\]]for the total probability remains equal to unity at all times . in the following
we omit this constant , as it is nothing but a scaling factor . in substituting the initial condition ( [ initial ] ) into eq.([aq0 ] ) and this into eq.([aprel ] ) , we are left with integrals of the form @xmath156whose solution can be expressed in terms of the airy function @xmath157 @xcite @xmath158 then the final result reads , up to a normalization factor , @xmath159or , explicitely , @xmath160with @xmath161 \\ & & \pm \sqrt{1-\rho } l_{0,0}\mathcal{z}\left ( \pm \left ( \xi + 1\right ) , \tau \right ) , \notag \\ l^{\pm } \left ( \xi , \tau \right ) & = & l_{0,0}\left [ \mathcal{z}\left ( \pm \xi , \tau \right ) \mp \sqrt{\rho } \mathcal{z}\left ( \pm \left ( \xi + 1\right ) , \tau \right ) \right ] \\ & & \pm \sqrt{1-\rho } r_{0,0}\mathcal{z}\left ( \pm \left ( \xi -1\right ) , \tau \right ) , \notag\end{aligned}\]]where use was made of the discrete qw eqs.([amp1 ] ) and ( [ amp2 ] ) . notice that while the form of eq .
( [ diff.eq . ] ) is independent of the particular form of @xmath162 , this solution depends explictly on @xmath151 because the application of the initial condition does depend on the details of that cut - off function .
finally , the probability distribution is given by @xmath163 let us now compare the above long wavelength approximation with the solution of the original discrete qw eqs.([amp1 ] ) and ( [ amp2 ] ) . in the left column of fig.2 we present the total probability distribution @xmath164 , as well as the partial probabilites @xmath165 and @xmath166 , as obtained from eqs.([amp1 ] ) and ( amp2 ) for @xmath167 , @xmath168 and @xmath169 ( a value of the initial conditions for which the mean displacement is null ) and @xmath170 ( which corresponds to the hadamard walk ) .
in the right column of fig.2 we present the solution of the long wavelength approximation @xmath171 ( and @xmath172and @xmath173 ) for @xmath174 and the same value of @xmath42 and of the initial conditions and for @xmath175 . finally , in the central column of the figure we present the long wavelength approximation evaluated only at integer values of the position @xmath141 .
these last plots are given in order to make a better comparison with the exact solution , as the latter is nonvanishing only at even ( odd ) positions at even ( odd ) times .
although we only show plots for a particular initial condition and value of @xmath42 , similar correspondence between the exact solution and the long wavelength limit results for other sets of initial conditions and values of @xmath42 as well . in fig .
2 we chose the value @xmath175 because for that particular value the agreement of the long wavelength approximation with the exact solution is close to the best that can be achieved . in fig .
3 we plot the long wavelength approximation for three different values of @xmath151 .
note that for the largest @xmath176 the long wavelength approximation yields values that are too low , compared to the exact solution , for small @xmath141 , while for the smallest @xmath177 the long wavelength approximation extends to @xmath178 beyond the range of the exact solution .
we see from fig .
2 that the main trends of the exact qw solution are well captured by the long wavelength approximation , although obvious quantitative differences appear .
the main discrepancy between the exact solution and the long wavelength approximation is the presence of the oscillations that appear in the latter . by taking into account higher order terms in the expansion ( [ series ] ) a more accurate approximation to the exact solution
could be obtained , but the corresponding differential equation would not have a simple analytic solution . in any case ,
our main thesis is not that eqs .
( [ solfin1 ] ) and ( [ solfin2 ] ) are a good quantitative approximation to the exact solution .
rather , our central result is that comparisons such as those in fig . 2
demonstrate that the propagation and dispersion effects described by eq .
( [ diff.eq . ] ) capture the qualitative nature of the qw .
in this paper we have rigorously derived both the exact solution of the coined qw for arbitrary unitary transformations , and a long wavelength approximation that constitutes an approximate continuous limit of the qw .
the long wavelength approximation identified here corresponds to the dynamical equation of an envelope function characterizing a light pulse propagating in a medium with higher order dispersion , for example in an optical fiber close to the zero - dispersion wavelength @xcite .
this , together with the fact that entirely classical implementations of the unidimensional qw are possible @xcite , lead us to the conclusion that entanglement does not play a central role in the unidimensional qw , and that the walk can be regarded as a pure wave phenomenon .
this conclusion can not , however , be generalized to higher dimensional qws @xcite .
there several coins are needed , and it would in such a scenario that the entanglement characteristic of an essential quantum nature might appear , as already discussed @xcite .
it would then be very interesting to generalize the work we have carried out here by deriving long wavelength descriptions of multidimensional qws .
this could help elucidate any role played by entanglement in such processes .
another point of interest is the effect of decoherence on qws .
kendon and tregenna @xcite have shown that decoherence modifies the probability distribution of the qw in somewhat unsuspected ways .
in particular , it can lead to quite smooth , and at the same time wide , probability distributions .
it would be interesting to investigate the role of decoherence in continuous descriptions of the qw , such as that we have presented here . in this sense , the recent work by romanelli _
@xcite , in which markovian and intereference terms of the quantum evolution are separated , provides a new interesting tool for this type of analysis .
this work has been supported in part by the uk engineering and physical sciences research council and the european union .
er acknowledges financial support from the ministerio de educacin , cultura y deportes of the spanish goverment ( grant pr20002 - 0244 ) .
jes acknowledges financial support from the natural sciences and engineering research council of canada .
we gratefully acknowledge fruitful discussions with v. kendon .
in this appendix we show the relation between the walk definition we use here and that used by nayak and vishwanath _ _ _ _ @xcite , and confirm that our solution reproduces theirs .
we begin with our walk in the case of a hadamard transformation , @xmath179and define new variables @xmath180so @xmath181 in terms of these new variables the walk eqs.([hadamard1 ] ) and ( hadamard2 ) are specified by @xmath182these are the walk equations of nayak and vishwanath @xcite__. _ _ then , in terms of our functions @xmath104 , where @xmath95 is @xmath96 or @xmath97 , we have from eqs.([finres ] ) and ( [ nfromus ] ) , @xmath183 in writing out the expressions for @xmath184 and @xmath185 at the lattice sites indicated we do a change of variables , putting @xmath186 .
we use @xmath187 & = & -\exp \left [ ik^{\prime } x% \right ] , \\
\exp \left [ -i(k^{\prime } + \pi /x)mx\right ] & = & ( -1)^{m}\exp \left [ -ik^{\prime } mx\right ] , \end{aligned}\]]and can then adjust the limits of integration of the function back to @xmath188 to @xmath189 in the new variable @xmath190 , since the function is periodic over @xmath191 . changing the name of the new variable back to @xmath61 we can collect the terms and write @xmath192 \notag \\ & & \cdot e^{-ikmx}e^{-i\omega _ { 0}(k)nt } , \\
\hat{l}_{m , n } & = & \int_{-\pi /x}^{\pi /x}dk\,e^{-ikx}\left [ r^{+}(k)-(-1)^{n+m}r^{-}(k+\frac{\pi } { x})\right ] \notag \\ & & \cdot e^{-ikmx}e^{-i\omega _ { 0}(k)nt}.\end{aligned}\]]now the initial condition adopted in _ _ _ _ @xcite is @xmath193 and @xmath194 , which , in our notation eqs.(usfromn ) read @xmath195 and @xmath196 after one time step , see eqs.([hadamard1 ] ) and ( [ hadamard2 ] ) , we have @xmath197 , so we can construct @xmath198 and @xmath199 from ( [ a(k ) ] ) , with @xmath10 .
we find @xmath200and using ( [ disp ] ) we recover @xmath201 \frac{x}{4\pi } \notag \\ & & \cdot \int_{-\pi /x}^{\pi /x}dk\frac{e^{ikx}}{\sqrt{1+\cos
^{2}kx}}% e^{-ikmx}e^{-i\omega _ { 0}(k)nt } , \\
\hat{l}_{m , n } & = & \left [ 1+(-1)^{n+m}\right ] \frac{x}{4\pi } \notag \\ & & \cdot \int_{-\pi /x}^{\pi /x}dk\left
( 1+\frac{\cos kx}{\sqrt{1+\cos ^{2}kx}% } \right ) e^{-ikmx}e^{-i\omega _ { 0}(k)nt},\end{aligned}\]]in agreement with nayak and vishwanath @xcite * fig.2 . * probability distribution of the qw .
the exact solution is presented in the left column ( @xmath202 , @xmath203 and @xmath166 for @xmath167 from top to bottom ) , the long wavelength approximation ( @xmath204 , @xmath205 , and @xmath206 for @xmath207 from top to bottom ) is presented in the right column , and also in the central column but in this case evaluated only at integer even . in the left and central columns the points have been joined to guide the eye .
the initial condition is @xmath168 and @xmath169 , and the parameters are @xmath10 , ( hadamard walk ) and @xmath175 for the long wavelength approximation .
the probabilities in the long wavelength approximation have not been normalized , and the units are thus arbitrary . | we analyze the solution of the coined quantum walk on a line .
first , we derive the full solution , for arbitrary unitary transformations , by using a new approach based on the four `` walk fields '' which we show determine the dynamics .
the particular way of deriving the solution allows a rigorous derivation of a long wavelength approximation .
this long wavelength approximation is useful as it provides an approximate analytical expression that captures the basics of the quantum walk and allows us to gain insight into the physics of the process . |
the recent solar @xcite , atmospheric @xcite , reactor @xcite and accelerator @xcite neutrino oscillation experiments have provided us with very robust evidence that neutrinos are massive and their flavor mixing involves two large angles ( @xmath11 and @xmath12 ) and one small angle ( @xmath13 ) .
these important results indicate that the minimal standard electroweak model , in which neutrinos are massless weyl particles , is incomplete
. a very economical extension of the standard model , the so - called minimal seesaw model @xcite , is to introduce two heavy right - handed majorana neutrinos ( @xmath14 and @xmath15 ) but preserve its @xmath16 gauge symmetry . in this case , the lagrangian relevant for lepton masses can be written as @xmath17 where @xmath18 denotes the left - handed lepton doublet , @xmath19 is the higgs - boson weak isodoublet ( @xmath20 ) , @xmath21 and @xmath22 stand respectively for the right - handed charged - lepton and majorana neutrino singlets . after spontaneous gauge symmetry breaking , we obtain the charged - lepton mass matrix @xmath23 and the dirac neutrino mass matrix @xmath24 with @xmath25 being the vacuum expectation value of @xmath19 . because @xmath14 and @xmath15 are @xmath26 singlets , their mass scale is not subject to the electroweak symmetry breaking and can be much higher than @xmath27
. then the seesaw mechanism @xcite works for the effective ( light and left - handed ) neutrino mass matrix @xmath28 ; i.e. , @xmath29 holds as a very good approximation .
the smallness of left - handed neutrino masses @xmath30 ( for @xmath31 ) is therefore ascribed to the largeness of right - handed neutrino masses @xmath32 ( for @xmath33 ) , while the bi - large neutrino mixing pattern is attributed to a significant mismatch between the diagonalizations of @xmath34 and @xmath28 .
note that @xmath35 allows the lepton - number - violating decays of @xmath36 ( for @xmath33 ) to happen : @xmath37 and @xmath38 .
since each decay mode can occur at both tree and one - loop levels , the interference of two decay amplitudes leads to a cp - violating asymmetry @xmath39 between @xmath40 and its cp - conjugated process @xmath41 . if the expansion of the universe gives rise to the out - of - equilibrium decays of @xmath14 and ( or ) @xmath15 @xcite , then @xmath42 and @xmath43 ( or one of them ) may result in a net lepton number asymmetry .
the latter can partially be converted into a net baryon number asymmetry via the nonperturbative sphaleron interaction @xcite .
such an elegant baryogenesis - via - leptogenesis mechanism @xcite provides a viable interpretation of the cosmological baryon number asymmetry , @xmath44 , which has recently been extracted from the three - year wmap observational data @xcite .
the most straightforward test of the seesaw and leptogenesis mechanisms is to discover the heavy right - handed majorana neutrinos and measure the strength of their interactions with other known particles , for instance , at the large hadron collider ( lhc ) and the international linear collider ( ilc ) @xcite in the near or foreseeable future .
nevertheless , @xmath36 can not be produced at lhc unless its mass scale @xmath32 is of @xmath0 tev up to @xmath45 tev , far below the typical seesaw scale @xmath46 gev @xcite .
if @xmath47 tev is taken , for example , a successful leptogenesis to account for the observed value of @xmath9 requires that the splitting between @xmath48 and @xmath49 be extremely small @xcite .
this interesting scenario , usually referred to as the resonant leptogenesis mechanism , may also get clear of the gravitino overproduction problem in those supersymmetric extensions of the standard model with heavy majorana neutrinos of @xmath50 .
but @xmath51 ev and @xmath47 tev in the seesaw relation imply that the corresponding yukawa couplings must be of @xmath52 or smaller , rendering the production and detection of @xmath36 rather dim at lhc and ilc @xcite .
for this reason , here we shall follow a less ambitious strategy by leaving out the observability of heavy right - handed majorana neutrinos and concentrating on the phenomenology of neutrino mixing and resonant leptogenesis .
the goal of this paper is to simultaneously explain the observed neutrino mixing pattern and the cosmological baryon number asymmetry .
we are going to propose a particularly economical and suggestive seesaw model with two highly degenerate right - handed majorana neutrinos of @xmath0 tev .
the novel yukawa - coupling texture of our model leads to the normal neutrino mass hierarchy with @xmath53 and a nearly tri - bimaximal neutrino mixing pattern with the maximal cp - violating phase : @xmath2 , @xmath3 and @xmath54 . on the other hand
, one may get the inverted neutrino mass hierarchy with @xmath55 and the corresponding neutrino mixing pattern with @xmath2 and @xmath56 .
we find that it is possible to establish a straightforward link between cp violation in the decays of heavy right - handed majorana neutrinos and that in the oscillations of light left - handed majorana neutrinos in the @xmath57 case . as for the cosmological baryon number asymmetry ,
we show that the observed value of @xmath9 can naturally be interpreted through the resonant leptogenesis mechanism in both @xmath53 and @xmath55 cases .
we point out that the formula of @xmath39 given in ref .
@xcite and that presented in ref .
@xcite lead to the same numerical result in the model under discussion , if the parameter space coincides with the perturbation condition .
more interestingly , we demonstrate the significance of flavor - dependent effects in our resonant leptogenesis scenario : they can either flip the sign of the flavor - independent prediction for @xmath9 in the @xmath58 case or magnify the magnitude of the flavor - independent prediction for @xmath9 about @xmath10 times in the @xmath59 case .
a salient feature of the minimal seesaw model is that it automatically predicts @xmath1 or @xmath5 @xcite .
hence the @xmath60 dirac neutrino mass matrix @xmath61 can be parameterized as @xmath62 with @xmath1 @xcite or as @xmath63 with @xmath5 , where @xmath64 and @xmath65 are @xmath66 and @xmath67 unitary matrices , respectively . without loss of generality
, we adopt the flavor basis in which both the charged - lepton mass matrix @xmath34 and the right - handed majorana neutrino mass matrix @xmath68 are diagonal , real and positive .
then the seesaw relation @xmath69 implies that the flavor mixing of light majorana neutrinos depends primarily on @xmath64 and the decays of heavy majorana neutrinos rely mainly on @xmath65 @xcite .
this phenomenological observation motivates us to take @xmath64 to be the tri - bimaximal mixing pattern @xcite @xmath70 which is compatible very well with the best fit of current experimental data on neutrino oscillations @xcite . on the other hand
, the unitary matrix @xmath65 can be parameterized as @xmath71 since @xmath72 is the only phase parameter in our model , it should be responsible both for the cp violation in neutrino oscillations and for the cp violation in @xmath36 decays . in order to implement the idea of resonant leptogenesis
, we assume that two heavy majorana neutrino masses are highly degenerate ; i.e. , the magnitude of @xmath73 is strongly suppressed .
indeed @xmath74 or smaller has typically been anticipated in some seesaw models with three right - handed majorana neutrinos @xcite to gain the successful resonant leptogenesis . given @xmath75 , the explicit form of @xmath28
can reliably be formulated from the seesaw relation @xmath29 by neglecting the tiny mass splitting between @xmath14 and @xmath15 .
in such a good approximation , we obtain @xmath76 \ ; % ( 5a ) \eqno{\rm ( 5a)}\ ] ] for @xmath1 ; or @xmath77 \ ; % ( 5b ) \eqno{\rm ( 5b)}\ ] ] for @xmath5 , where @xmath78 is defined .
one may follow the procedure outlined in ref .
@xcite to diagonalize the effective neutrino mass matrix in eq .
( 5a ) or ( 5b ) . for simplicity
, here we fix @xmath79 and highlight the role of @xmath72 in neutrino mixing and leptogenesis .
then both the neutrino mass spectrum and the neutrino mixing pattern can be achieved from the simplified version of @xmath80 or @xmath81 . * normal neutrino mass hierarchy with @xmath53 . in this case
, we diagonalize @xmath80 by using the transformation @xmath82 , where @xmath83 is just the neutrino mixing matrix . after a straightforward calculation , we have @xmath84 \ ; , \\ m^{}_3 & = & \displaystyle \frac{y^2}{2m^{}_2 } \left [ \sqrt{\left ( 1 + \omega^2 \right ) ^2
\cos^2 2\alpha + 4 \omega^2 \sin^2 2\alpha } ~ + \left ( 1 - \omega^2 \right ) \left
| \cos 2\alpha \right | \right ] \ ; , \end{array } % ( 6a ) \eqno{\rm ( 6a)}\ ] ] where @xmath85 .
taking account of @xmath86 and @xmath87 , we obtain @xmath88 ev and @xmath89 ev by using @xmath90 and @xmath91 @xcite as the typical inputs . furthermore , @xmath92 where @xmath93 is given by @xmath94 . comparing this result with the standard parametrization of @xmath83 @xcite
, we immediately arrive at @xmath95 and vanishing majorana phases of cp violation .
( 8a ) implies a very interesting correlation between @xmath96 and @xmath97 : @xmath54 . when @xmath98 , we get @xmath99 , which is almost the best - fit value of the solar neutrino mixing angle @xcite ( i.e. , the large - mixing - angle msw solution to the solar neutrino problem @xcite ) .
note that the smallness of @xmath97 requires the smallness of @xmath93 , or equivalently the smallness of @xmath72 .
( 6a ) and ( 8a ) , together with @xmath100 and the values of @xmath101 and @xmath102 obtained above , yield @xmath103 , @xmath104 and @xmath105 .
( 6a ) can reliably approximate to @xmath106 and @xmath107 for @xmath108 .
the jarlskog rephasing - invariant parameter @xmath109 @xcite , which determines the strength of cp violation in neutrino oscillations , is found to be @xmath110 in this scenario .
it is possible to measure @xmath111 in the future long - baseline neutrino oscillation experiments . * inverted neutrino mass hierarchy with @xmath55 . in this case , we diagonalize @xmath81 by using the transformation @xmath112
. then we obtain @xmath113 \ ; , \\ m^{}_2 & = & \displaystyle \frac{y^2}{2m^{}_2 } \left [ \sqrt{\left ( 1 + \omega^2 \right ) ^2 \cos^2 2\alpha + 4 \omega^2 \sin^2 2\alpha } ~ + \left ( 1 - \omega^2 \right ) \left
| \cos 2\alpha \right | \right ] \ ; , \end{array } % ( 6b ) \eqno{\rm ( 6b)}\ ] ] where @xmath85 .
taking account of @xmath114 and @xmath115 , we get @xmath116 ev and @xmath117 ev by inputting @xmath90 and @xmath91 . in addition , @xmath118 where @xmath93 is given by @xmath119 . when the standard parametrization of @xmath83 @xcite is applied to eq .
( 7b ) , we have @xmath120 and vanishing majorana phases of cp violation . eq .
( 8b ) indicates that @xmath81 actually has the @xmath121-@xmath122 symmetry and cp is conserving at low - energy scales in this scenario .
given @xmath123 @xcite , @xmath93 is found to lie in the range @xmath124 .
this result , together with eq .
( 6b ) and the values of @xmath125 and @xmath101 obtained above , allows us to get @xmath126 and @xmath127 .
we observe that @xmath128 and @xmath129 are good approximations of eq .
( 6b ) for @xmath108 . from a phenomenological point of view ,
the scenario with @xmath53 is more favored and more interesting than the scenario with @xmath130 .
both of them can be tested in the near future .
now let us consider the lepton - number - violating and cp - violating decays of two heavy right - handed majorana neutrinos .
their decay widths are given by @xmath131 at the tree level , where @xmath132 .
with the help of eq .
( 2 ) , it is easy to verify that @xmath133 has a universal form for both @xmath53 and @xmath55 cases : @xmath134 given the near degeneracy between @xmath48 and @xmath49 as well as @xmath79 for @xmath65 , it turns out that @xmath135 is an excellent approximation .
but the decay mode @xmath37 and its cp - conjugated process @xmath136 can actually occur at both tree and one - loop levels .
hence their cp - violating asymmetry @xmath39 , defined as the ratio of @xmath137 to @xmath138 , arises from the interference of two decay amplitudes @xcite .
note that the self - energy correction dominates over the vertex correction at the one - loop level , as the former is resonantly enhanced by the tiny mass splitting between @xmath14 and @xmath15 .
in other words , @xmath39 primarily results from the interference between the tree - level amplitude and its self - energy correction @xcite , decays .
this flavor - independent leptogenesis scenario actually serves as a counter example to illustrate why flavor effects are important and must be taken into account in a tev - scale resonant leptogenesis model , as one can clearly see in section iv .
] @xmath139}{\left ( y^\dagger_\nu y^{}_\nu \right ) ^{}_{11 } \left ( y^\dagger_\nu y^{}_\nu \right ) ^{}_{22 } } \cdot \frac{\left ( m^2_i - m^2_j \right ) m^{}_i \gamma^{}_{j}}{\left ( m^2_i - m^2_j \right ) ^2 + m^2_i \gamma^{2}_{j } } \;\ ; , % ( 10)\ ] ] where @xmath140 and @xmath141 run over 1 and 2 but @xmath142 .
combining eqs .
( 9 ) and ( 10 ) , we obtain the explicit expression of @xmath39 in our model : @xmath143 } ~ r \sin 4\alpha \ ; , % ( 11)\ ] ] in which @xmath144 has been defined to describe the mass splitting between two heavy majorana neutrinos .
of course , @xmath145 is also an excellent approximation due to @xmath146 .
these two cp - violating asymmetries would vanish if @xmath147 or @xmath148 were taken .
to estimate the order of @xmath39 at the tev scale , we restrict ourselves to the interesting @xmath108 region and make use of the approximate result @xmath149 ( or @xmath150 ) obtained above for the @xmath53 ( or @xmath55 ) case . * in the @xmath53 case , we get @xmath151 from @xmath89 ev and @xmath152 tev .
in addition , @xmath153 .
then eq . (
11 ) is approximately simplified to @xmath154 together with @xmath155 for @xmath156 .
note that @xmath157 is in general expected so as to achieve the successful leptogenesis ( see below ) .
hence the third possibility @xmath158 requires @xmath159 , implying very tiny ( unobservable ) cp violation in neutrino oscillations .
if @xmath160 , one may take either @xmath161 or @xmath162 to obtain @xmath163 .
* in the @xmath55 case , we obtain @xmath164 from @xmath117 ev and @xmath152 tev . in addition , @xmath165 .
then eq . (
11 ) is approximately simplified to @xmath166 together with @xmath167 for @xmath168 . given either @xmath169 or @xmath170
, it is straightforward to arrive at @xmath171 for @xmath160 .
if @xmath172 holds , on the other hand , @xmath173 will be impossible to reach @xmath174 .
note again that @xmath175 has been obtained in eq .
( 8b ) , implying that @xmath176 is completely independent of @xmath175 in this scenario .
as described before , the cp - violating asymmetries @xmath42 and @xmath43 can give rise to a net lepton number asymmetry in the universe , provided its expansion rate is larger than @xmath177 and @xmath178 .
the nonperturbative sphaleron interaction may partially convert this lepton number asymmetry into a net baryon number asymmetry @xcite , @xmath179 where @xmath180 and @xmath181 are the efficiency factors measuring the washout effects associated with the out - of - equilibrium decays of @xmath14 and @xmath15 , and @xmath182 is an excellent approximation in our model . to evaluate the magnitude of @xmath183 ,
let us take account of the effective neutrino masses @xmath184 .
of course , @xmath185 is also an excellent approximation . given @xmath108 , @xmath186 holds for the normal neutrino mass hierarchy ( @xmath53 ) ; and @xmath187 holds for the inverted neutrino mass hierarchy ( @xmath55 ) .
when @xmath188 , the washout of the baryon number asymmetry is so effective that the final asymmetry can hardly depend on the initial one .
alternatively , one may define the parameters @xmath189 at @xmath190 , where @xmath191 denotes the temperature of the universe , @xmath192 is the hubble constant with @xmath193 , and @xmath194 represents the planck mass .
the relationship between @xmath195 and @xmath196 is rather simple : @xmath197 , where @xmath198 is the so - called equilibrium neutrino mass . for simplicity
, the efficiency factors @xmath183 can be estimated by using the approximate formula @xcite @xmath199 which is valid when the masses of two heavy right - handed majorana neutrinos are nearly degenerate and the parameters @xmath195 lie in the range @xmath200 . with the help of eq .
( 14 ) , we get @xmath201 and @xmath202 for the @xmath53 case ; or @xmath203 and @xmath204 for the @xmath55 case .
finally , the cosmological baryon number asymmetry is found to be @xmath205 in the @xmath1 case ; or @xmath206 in the @xmath5 case .
note that the numerical results in eq .
( 16 ) are obtained by taking @xmath152 tev .
other numerical results can similarly be achieved by starting from eq . ( 11 ) and allowing @xmath49 to vary , for instance , from 1 tev to 10 tev .
it is certainly possible to get @xmath8 from eq .
( 16 ) with a proper choice of @xmath207 and @xmath72 , either in the @xmath53 case or in the @xmath55 case .
but one should keep in mind that the flavor - independent leptogenesis scenario under discussion has a potential problem , because it ignores the fact that all the yukawa interactions of charged leptons are in thermal equilibrium at temperatures of @xmath0 tev @xcite .
we shall consider the flavor - dependent effect in the next section and demonstrate its significance in our resonant leptogenesis model .
we remark that the formula of @xmath39 in eq .
( 10 ) is quoted from ref .
a different formula has recently been presented in ref .
@xcite to describe the resonant cp - violating asymmetry for two nearly degenerate right - handed majorana neutrinos : @xmath208}{\left ( y^\dagger_\nu y^{}_\nu \right ) ^{}_{11 } \left ( y^\dagger_\nu y^{}_\nu \right ) ^{}_{22 } } \cdot \frac{\left ( m^2_i - m^2_j \right ) m^{}_i \gamma^{}_{j}}{\left ( m^2_i - m^2_j \right ) ^2 + \left(m^{}_j \gamma^{}_{j } - m^{}_i \gamma^{}_{i}\right)^2 } \;\ ; , % ( 17)\end{aligned}\ ] ] where the perturbation condition @xmath209 $ ] should be satisfied .
which formula is more reliable remains controversial at present . here
let us compare between the results of @xmath39 and @xmath210 in our model .
taking account of @xmath211 where @xmath212 for @xmath142 , we obtain the ratio @xmath213 from eqs .
( 10 ) and ( 17 ) as an excellent approximation .
note that @xmath214 ( for @xmath1 ) or @xmath215 ( for @xmath5 ) in the numerical examples taken above .
note also that the perturbation condition used in deriving eq .
( 17 ) is equivalent to @xmath216 .
hence we arrive at @xmath217 when @xmath218 holds , or equivalently @xmath219 is satisfied . in other words , the formula of @xmath39 given in ref .
@xcite and that presented in ref .
@xcite lead to the same numerical results in our leptogenesis scenario , if and only if the parameter space coincides with the perturbation condition .
this numerical agreement is by no means a hint that there might not exist any serious discrepancy between refs .
@xcite and @xcite .
it is actually desirable to understand the properties of unstable particles in quantum field theories to the utmost extent , so as to clarify the potential ambiguities associated with @xmath39 and @xmath210 ( or one of them ) in the resonant leptogenesis models .
flavor - dependent effects in leptogenesis have recently attracted a lot of attention @xcite . since all the yukawa interactions of charged leptons are in thermal equilibrium at the tev scale , the flavor issues of @xmath36 decays should be taken into account in our model . in the framework of resonant leptogenesis , it is straightforward to calculate the cp - violating asymmetry between @xmath220 and @xmath221 decays for each lepton flavor @xmath72 ( @xmath222 , @xmath121 or @xmath122 ) : @xmath223 } \nonumber \\ & = & \frac{8\pi \left(m^2_i - m^2_j\right ) { \rm im}\left\ { \left(y^{}_\nu\right)^{}_{\alpha j } \left(y^{}_{\nu } \right)^{*}_{\alpha i } m^{}_i \left[m^{}_i \left(y^\dagger_\nu y^{}_\nu \right)^{}_{ji } + m^{}_j \left(y^\dagger_\nu y^{}_\nu \right)^{}_{ij } \right ]
\right\}}{\left [ 64\pi^2 \left(m^2_i -m^2_j\right)^2 + m^4_i \left(y^\dagger_\nu y_\nu \right)^{2}_{jj } \right]\left(y^\dagger_\nu y_\nu \right)^{}_{ii } } \ ; , % ( 20)\end{aligned}\ ] ] where @xmath140 and @xmath141 run over @xmath224 and @xmath225 but @xmath226 . combining eqs .
( 2 ) and ( 20 ) , we explicitly obtain @xmath227 in the @xmath1 case ; and @xmath228 in the @xmath5 case , where @xmath39 has been given in eq . ( 11 ) . because @xmath145 is an excellent approximation , one can see from eq .
( 21 ) that @xmath229 is also an excellent approximation . in view of @xmath230 for @xmath53 or @xmath127 for @xmath5
, we have @xmath231 and @xmath232 in both cases .
the sum of these three flavor - dependent asymmetries is just the total cp - violating asymmetry @xmath39 ; i.e. , @xmath233 holds exactly .
once the initial values of @xmath234 are fixed , the final result of @xmath9 will be governed by a set of flavor - dependent boltzmann equations including the ( inverse ) decay and scattering processes as well as the nonperturbative sphaleron interaction @xcite . in order to estimate the washout effects , one may introduce the parameters @xmath235 } k^{}_i = \frac{\left| ( y^{}_\nu)^{}_{\alpha i } \right|^2 } { \left(y^\dagger_\nu y_\nu \right)^{}_{ii } } k^{}_i \ ; , % ( 22)\end{aligned}\ ] ] in which @xmath236 at @xmath237 has been defined before .
of course , @xmath238 holds . with the help of eq .
( 2 ) , we get either @xmath239 in the @xmath1 case ; or @xmath240 in the @xmath5 case .
the flavor - dependent effects in leptogenesis with almost degenerate heavy majorana neutrinos have been elucidated in ref .
@xcite , where a useful approximation for the efficiency factor of each flavor is given as @xmath241 \ ; % ( 24)\end{aligned}\ ] ] with @xmath242 and @xmath243 .
the above expression can well approximate to the power law @xmath244 for @xmath245 . recalling @xmath246 ( or @xmath127 ) and @xmath201 ( or @xmath203 ) for @xmath53 ( or @xmath55 ) , one can then evaluate the decay parameters @xmath247 : @xmath248 and @xmath249 ( or @xmath250 and @xmath251 ) .
the final baryon number asymmetry can be calculated via @xmath252 where the superscript means that flavor effects have been included ; @xmath253 and @xmath254 are given in eqs .
( 21 ) and ( 24 ) , respectively . a straightforward comparison between @xmath9 in eq .
( 13 ) and @xmath255 in eq .
( 25 ) can clearly demonstrate the significance of flavor effects in our resonant leptogenesis scenario : @xmath256 note that flavor effects flip the sign of @xmath9 in the @xmath257 case .
this nontrivial result can be understood as follows .
the novel yukawa structure of our model dictates the electron flavor to make a negative contribution to the cp - violating asymmetry ( @xmath258 ) and to have a large efficiency factor ( due to the smallness of @xmath259 ; i.e. , @xmath260 ) .
although the other two flavors contribute significantly to the cp - violating asymmetry , their corresponding washout effects are much stronger ( @xmath261 ) .
hence the overall effects of three flavors in @xmath262 make its sign opposite to that of @xmath9 in the @xmath53 case . in the @xmath55 case
, the large and negative contribution of the electron flavor to the cp - violating asymmetry is badly diluted by its extremely small efficiency factor ( due to the largeness of @xmath259 ; i.e. , @xmath250 ) , and thus the muon and tau flavors make the dominant contributions to @xmath255 via both @xmath263 and @xmath264 .
that is why @xmath255 has the same sign as @xmath9 in the @xmath55 case , in which the large ratio @xmath265 is essentially ascribed to the large enhancement factors @xmath266 and @xmath267 . combining eqs .
( 16 ) and ( 26 ) , we are then able to obtain the observed value of @xmath255 via the flavor - dependent tev - scale leptogenesis with proper inputs of @xmath207 and @xmath72 . as an example , fig .
1 illustrates the numerical correlation between @xmath207 and @xmath72 to get @xmath268 in both @xmath53 and @xmath269 cases , where @xmath270 tev , 2 tev , 3 tev , 4 tev and 5 tev have typically been input and eq . ( 11 ) has also been used .
note that we only concentrate on the region @xmath271 , which satisfies the perturbation condition given in ref .
one may clearly see the similar behavior of @xmath207 changing with @xmath72 in both cases , in which @xmath272 holds as the leading - order approximation . in order to get the correct sign for the cosmological baryon number asymmetry , however
, we should take @xmath273 in the @xmath257 case . since negative @xmath72 leads to negative @xmath93 in eq .
( 7a ) , the sign of @xmath274 in eq .
( 8a ) will flip . but this modification does not change the results of neutrino masses and flavor mixing angles obtained in eqs .
( 6a ) and ( 8a ) .
in summary , we have proposed a very simple but suggestive seesaw model with two highly degenerate right - handed majorana neutrinos of @xmath0 tev .
its novel yukawa - coupling texture leads to the normal neutrino mass hierarchy with @xmath53 and a nearly tri - bimaximal neutrino mixing pattern with the maximal cp - violating phase : @xmath2 , @xmath3 and @xmath275 . on the other hand
, it is possible to get the inverted neutrino mass hierarchy with @xmath55 and the corresponding neutrino mixing pattern with @xmath2 and @xmath56 . a straightforward link between cp violation in the decays of heavy right - handed majorana neutrinos and that in the oscillations of light left - handed majorana neutrinos can be established in the @xmath53 case .
we have successfully interpreted the cosmological matter - antimatter asymmetry @xmath276 through the flavor - dependent resonant leptogenesis mechanism in both @xmath53 and @xmath55 cases . in particular , we have shown that flavor effects can either flip the sign of the flavor - independent prediction for @xmath9 in the @xmath58 case or magnify the magnitude of the flavor - independent prediction for @xmath9 about 50 times in the @xmath5 case .
* we have made three assumptions in building the model .
taking @xmath64 to be the tri - bimaximal mixing pattern is purely a phenomenological assumption @xcite , so is taking @xmath65 to be the maximal mixing pattern with a cp - violating phase . both of them might be able to result from certain flavor symmetries ( e.g. , discrete @xmath277 and @xmath278 symmetries @xcite ) .
the third assumption is the near degeneracy between @xmath48 and @xmath49 , or equivalently the smallness of @xmath207 .
it seems a bit contrived to take such tiny values of @xmath207 , as shown in fig . 1 , to achieve the successful baryogenesis via leptogenesis .
why is @xmath207 so small ?
a few authors @xcite have conjectured that @xmath48 and @xmath49 might be exactly degenerate at an energy scale higher than the seesaw scale ( e.g. , the gut scale ) and their small splitting at the seesaw scale is attributed to the radiative corrections .
* the minimal seesaw model with two heavy right - handed majorana neutrinos @xmath14 and @xmath15 can be regarded as a limiting version of the conventional seesaw model with three right - handed majorana neutrinos @xmath14 , @xmath15 and @xmath279 , in which @xmath279 is so heavy that it decouples from the theory at the very early stage . in order to gain the special mass spectrum ( i.e. , @xmath280 )
, one possibility is to make use of the froggatt - nielsen ( fn ) mechanism and assign the charges of three right - handed majorana neutrinos as @xmath281 , @xmath282 and @xmath283 @xcite .
such a scenario allows @xmath14 and @xmath15 to be of @xmath0 tev and @xmath279 to be close to the gut scale , and it can also accommodate the yukawa couplings of @xmath284 .
besides flavor symmetries , the supersymmetry breaking may do the same job , as shown in ref .
radiative corrections to the light neutrino masses , the mixing angles and the cp - violating phases should also be taken into account , when those parameters run from the seesaw scale down to the electroweak scale .
a generic analysis of the renormalization - group running effects in the minimal seesaw model has been done in ref .
@xcite . because @xmath285 tev is specified in our model , however ,
all of such running effects are expected to be negligibly small . even in the supersymmetric case with large @xmath286
, significant radiative corrections can not emerge between the energy scales of @xmath0 tev and the electroweak scale . *
we stress that this tev - scale neutrino mass model is also viable in the framework of the minimal supersymmetric standard model .
in particular , the above calculation of @xmath9 can simply be extended to accommodate supersymmetry .
the size of @xmath39 in the supersymmetric case is twice as large as in the standard model , so is the magnitude of @xmath287 @xcite .
the enhancement of @xmath287 gives rise to the same order suppression of the @xmath288 dilution coefficient in eq .
( 13 ) or eq .
( 25 ) , hence the estimate of @xmath9 is not significantly changed by the introduction of supersymmetry .
this observation , which has been confirmed by our detailed analysis , implies that the numerical results shown in fig . 1 are expected to be roughly ( order - of - magnitude ) valid for the minimal supersymmetric standard model .
the upcoming neutrino oscillation experiments will test the predictions of our model for the neutrino mass spectrum , flavor mixing angles and cp violation .
a more intriguing test of such seesaw - plus - leptogenesis models is certainly the direct search for tev - scale right - handed majorana neutrinos , are strongly model - dependent .
for example , they inevitably rely on the mixing strength between @xmath36 and three charged leptons . the latter has been constrained to some extent by current experimental data on the lepton universality and lepton - flavor - violating processes .
@xcite for detailed discussions . ] which can be done at lhc and ilc in the future .
due to the tiny yukawa couplings restricted by the models themselves , however , the observability of such new particles will be a big experimental challenge .
p. minkowski , phys .
b * 67 * , 421 ( 1977 ) ; t. yanagida , in _ proceedings of the workshop on unified theory and the baryon number of the universe _ , edited by o. sawada and a. sugamoto ( kek , tsukuba , 1979 ) , p. 95 ; m. gell - mann , p. ramond , and r. slansky , in _ supergravity _ , edited by p. van nieuwenhuizen and d. freedman ( north holland , amsterdam , 1979 ) , p. 315 ; s.l . glashow , in _ quarks and leptons _ , edited by m. l@xmath289 _ et al .
_ ( plenum , new york , 1980 ) , p. 707 ; r.n .
mohapatra and g. senjanovic , phys .
lett . * 44 * , 912 ( 1980 ) .
it seems possible to get around this difficulty by assuming a very contrived flavor structure in the resonant leptogenesis scenario .
see , e.g. , a. pilaftsis , phys .
lett . * 95 * , 081602 ( 2005 ) ; a. pilaftsis and t.e.j .
underwood , phys .
d * 72 * , 113001 ( 2005 ) .
harrison , d.h .
perkins , and w.g .
scott , phys .
b * 530 * , 167 ( 2002 ) ; z.z .
xing , phys .
b * 533 * , 85 ( 2002 ) ; p.f .
harrison and w.g .
scott , phys .
b * 535 * , 163 ( 2002 ) ; x.g .
he and a. zee , phys .
b * 560 * , 87 ( 2003 ) .
a. abada , s. davidson , f.x .
josse - michaux , m. losada , and a. riotto , jcap * 0604 * , 004 ( 2006 ) ; e. nardi , y. nir , e. roulet , and j. racker , jhep * 0601 * , 164 ( 2006 ) ; a. abada , s. davidson , f.x .
josse - michaux , m. losada , and a. riotto , hep - ph/0605281 ; s. antusch , s.f .
king , and a. riotto , hep - ph/0609038 ; s. pascoli and s.t .
petcov , hep - ph/0609125 ; g.c .
branco , r. gonzalez felipe , and f.r .
joaquim , hep - ph/0609297 ; s. blanchet , p. di bari , g.g .
raffelt , hep - ph/0611337 ; s. pascoli , s.t .
petcov and a. riotto , hep - ph/0611338 ; a. de simone and a. riotto , hep - ph/0611357 ; s. antusch , arxiv:0704.1591 [ hep - ph ] .
k. turzynski , phys .
b * 589 * , 135 ( 2004 ) ; r. gonzale felipe , f.r .
joaquim , and b.m .
nobre , phys .
d * 70 * , 085009 ( 2004 ) ; g.c .
branco , r. gonzalez felipe , f.r .
joaquim , and b.m .
nobre , phys .
b * 633 * , 336 ( 2006 ) . | we propose a particularly economical neutrino mass model , in which there are only two right - handed majorana neutrinos of @xmath0 tev and their masses are highly degenerate .
its novel yukawa - coupling texture together with the seesaw mechanism allows us to achieve the normal neutrino mass hierarchy with @xmath1 and a nearly tri - bimaximal neutrino mixing pattern with the maximal cp - violating phase : @xmath2 , @xmath3 and @xmath4 .
one may also obtain the inverted neutrino mass hierarchy with @xmath5 and the corresponding neutrino mixing pattern with @xmath6 and @xmath7 . in both cases , it is possible to interpret the cosmological baryon number asymmetry @xmath8 through the resonant leptogenesis mechanism .
we demonstrate the significance of flavor - dependent effects in this leptogenesis scenario : they can either flip the sign of the flavor - independent prediction for @xmath9 in the @xmath1 case or magnify the magnitude of the flavor - independent prediction for @xmath9 about @xmath10 times in the @xmath5 case . |
the theory of cosmic inflation @xcite , advanced to solve the cosmological horizon and flatness problems , also explains the initial fluctuations which later gave rise to galaxies and large - scale structure in the universe . while inflation generically predicts initial matter density fluctuations with an approximately gaussian random distribution , the small deviations from gaussianity that characterize different inflation models
have been suggested to provide an observational probe to test and distinguish the models .
while purely gaussian initial density fluctuations are fully described by their power spectrum , primordial non - gaussianity ( png ) requires higher - order statistics to characterize it , the lowest - order being the 3-point correlation function , or its fourier transform the bispectrum which is zero for the gaussian case .
henceforth , we will describe the level of png predicted by different inflation models in terms of this bispectrum as parametrized by the dimensionless `` nonlinearity parameter '' @xmath3 , specialized to the case of the so - called `` local '' template .
[ for further details , see @xcite and refs . therein . ] the standard simplest model
slow roll , single - field inflation predicts an extremely small level of png , given by @xmath4 , where @xmath5 is the spectral index of the primordial power spectrum @xcite .
recent cosmic microwave background ( cmb ) temperature anisotropy measurements find @xmath6 @xcite , so @xmath7 .
other more general models ( e.g. multi - field inflation ) predict much larger values of @xmath3 .
observational cosmology has made important progress in constraining png thus far .
recent measurement of the cmb anisotropy bispectrum by _ planck _ has placed the most stringent constraint so far , @xmath8 @xcite . since
, even in the ideal noise - free limit , cmb temperature ( temperature+polarization ) measurements can only reduce the error to @xmath9 @xcite , there is great interest in finding other methods to measure png ; if future observations still do not detect png , an error budget @xmath10 will be necessary to rule out non - standard inflationary models conclusively .
[ henceforth , we focus on this `` local '' template and remove the label `` local '' . ]
png affects the clustering of the early star - forming galactic halos responsible for creating a network of ionized patches in the surrounding intergalactic medium ( igm ) during the epoch of reionization ( eor ) , which leaves a png imprint on the tomographic mapping of neutral hydrogen in the igm using its redshifted 21 cm radiation .
we shall here investigate in detail the prospects for constraining png with radio interferometric 21 cm measurements .
our method and results differ significantly from previous attempts in the literature @xcite , as follows : ( 1 ) we apply the ionized density bias derived by ref .
@xcite to model the effect of png by the excursion - set model of reionization ( esmr ) ; ( 2 ) we show a phenomenological model that can constrain png just as accurately , independent of reionization details ; ( 3 ) we show that a single - epoch measurement can be tuned to the optimum frequency for constraining png ; ( 4 ) we show how combining multi - epoch measurements further reduces the forecast errors .
the 3-d power spectrum of 21 cm brightness temperature fluctuations ( hereafter , `` 21 cm power spectrum '' ) in observer s redshift space can be expressed to linear order in _
neutral _ and _ total _ hydrogen density fluctuations , @xmath11 and @xmath12 , respectively , as the sum of powers of @xmath13 ( cosine of angle between line - of - sight ( los ) @xmath14 and wave vector @xmath15 of a given fourier mode ) @xcite , @xmath16 @xmath17 $ ] , where @xmath18 @xmath19 , and @xmath20 is the global neutral fraction .
@xmath21 is the power spectrum between fields @xmath22 and @xmath23 . here
, we focus on the limit where spin temperature @xmath24 , valid soon after reionization begins . as such
, we can neglect the dependence on spin temperature , but our discussion can be readily generalized to finite @xmath25 .
we also focus on the 21 cm signal on large scales @xmath26 , so that linearity conditions are met ( see ref .
@xcite for a summary of these conditions ) . when the typical size of ionized regions is much smaller than the scale of interest
, nonlinear effects of reionization patchiness on the 21 cm power spectrum @xcite can be neglected . if we define _ neutral _ and _ ionized _ density biases , @xmath27 and @xmath28 , according to @xmath29 , i.e. ratio of density fluctuation in field @xmath22 to that of total matter density in fourier space , then the 21 cm power spectrum can be rewritten as @xmath30 ^ 2\ , p_{\delta\delta}(k , z)\ , , \label{eqn:21power}\ ] ] where @xmath31 is the total matter density power spectrum . here
, we assume the baryon distribution traces the cold dark matter on large scales , so @xmath32 .
the ionized density bias is the fundamental quantity derived from reionization models , related to the neutral density bias by @xmath33 where @xmath34 .
we model reionization with png , as follows , based on the results of ref .
@xcite . ,
using esmr with the values of @xmath35 marked in the legend.,width=325 ] \(i ) esmr : the basic postulate of esmr@xcite is that the local ionized fraction within a spherical volume with radius @xmath36 is proportional to the local collapsed fraction of mass in luminous sources above some mass threshold @xmath37 , i.e. @xmath38 , where @xmath39 parametrizes the efficiency of this mass in releasing ionizing photons into the igm . for simplicity , we assume atomic - cooling halos ( achs ) , i.e. halos with virial temperature @xmath40 , are the only sources of ionizing radiation . our methodology for esmr with png is as follows .
the collapsed fraction of achs in ref .
@xcite ( see also @xcite ) , calculated in the non - markovian extension to the excursion set formalism @xcite , for a given png parameter , is applied to the esmr formalism to calculate the ionized density bias , for a given efficiency , analytically , as described in detail in ref .
henceforth , since the functions @xmath41 and @xmath42 are set by two parameters @xmath43 , given a fiducial cosmology , so is the 21 cm power spectrum @xmath44 at any @xmath45 . as figure [ fig:21power ] illustrates , while @xmath39 changes the amplitude of the 21 cm power spectrum , @xmath0 changes the shape at small @xmath46 significantly .
[ note that the reionization history is virtually independent of @xmath0 for @xmath47 , e.g. , for @xmath48 , @xmath49 at @xmath50 for @xmath51 , respectively . ]
the ( nonzero ) minimum of the curve for @xmath52 is at wavenumber @xmath53 , where @xmath54 .
( ii ) phenomenological ( `` pheno''- ) model : just as png exhibits a scale - dependent effect on halo bias @xcite , so we also find , in ref .
@xcite , a scale - dependent non - gaussian correction to the ionized density bias , @xmath55 , where @xmath56 is the gaussian ionized density bias and scale - independent .
[ there is also a scale - independent non - gaussian correction , @xmath57 .
however , @xmath58 for @xmath59 ( see @xcite ) , so we neglect it here , similar to the neglect of a scale - independent non - gaussian correction to the halo bias when constraining png with galaxy surveys @xcite . ] for the local template , we derived from the esmr a relation between @xmath60 and @xmath56 in @xcite . on large scales ,
@xmath61 \frac{\delta_c \omega_m ( h_0/c)^2}{g(0 ) d(z)k^2 t(k ) } , \label{eqn : sdbias}\ ] ] where @xmath62 is the critical density in the spherical collapse model ( in an einstein - de sitter universe ) ; @xmath63 is the linear growth factor normalized to unity at @xmath64 ; @xmath65 in our fiducial cosmology , where @xmath66 corresponds to the initial epoch , i.e. limit of large redshift ; and @xmath67 is the matter transfer function normalized to unity on large scales .
this relation was further tested and confirmed by numerical solution of the linear perturbation theory of reionization ( lptr ) which includes radiative transfer @xcite .
[ for further details , see ref .
@xcite . ] henceforth , eq . ( [ eqn : sdbias ] ) is assumed to be _ generic _ , regardless of reionization details . in what we call the `` pheno - model '' , the 21 cm power spectrum
is set by three parameters , @xmath0 , @xmath68 and @xmath56 , at a given redshift .
the latter two parameters embrace our ignorance of reionization .
.specifications for 21 cm interferometers .
we assume observation time @xmath69 for each redshift bin of bandwidth @xmath70.[tab : spec ] [ cols="^,^,^,^,^,^",options="header " , ] [ tab:1sigall ] \(ii ) multi - epoch constraints : while a futuristic single epoch measurement can achieve a remarkable accuracy of @xmath71 ( 16 ) for omniscope ( ska ) , adding tomographic information can further improve the accuracy . the pheno - model alone can not be used to combine multi - frequency measurements because it does not specify the redshift evolutions of @xmath68 and @xmath56 . on the other hand , a model such as the esmr
can be used to combine multi - frequency measurements because it fixes the reionization history ( and therefore @xmath72 ) for a given ( @xmath0 , @xmath39 ) .
we show multi - epoch constraints from the esmr in table [ tab:1sigall ] .
specifically , if information is combined from @xmath73 ( 7-band , total 42 mhz bandwidth , corresponding to @xmath74 in the esmr ) , the constraint can be significantly tightened , i.e. @xmath75 for omniscope ( 10 times smaller than an ideal cmb experiment ) , and @xmath76 by ska ( two times smaller than _ planck _ ) .
a prior of @xmath77 from _ planck_+_wmap _
cmb measurements@xcite corresponds to a prior of @xmath78 , much larger than allowed by ska and omniscope alone , so adding this @xmath79 prior can not improve @xmath80 from these experiments .
if we take a more conservative upper limit , @xmath81 ( instead of @xmath82 as assumed above ) , then @xmath80 for omniscope is @xmath83 times larger . since multi - epoch observations tighten @xmath0 constraints by combining information from different frequency bands to increase the amount of data relative to a single band , our use of the simple esmr model here , with constant efficiency parameter @xmath39 , for which reionization spans a relatively narrow range of redshift , may be a conservative one .
if reionization is more extended , as in self - regulated reionization models@xcite , for example , the resulting @xmath0 constraints may be even tighter .
previously , ref .
@xcite reported forecasts of @xmath84 $ ] for [ mwa512 , lofar , ska , omniscope ] based on information from a single redshift bin at the 50%-ionized epoch .
their results can be compared to the bottom set in table [ tab:1sig25 ] .
our results differ for a number of reasons : ( 1 ) ref.@xcite computed the scale - dependent signature of png in @xmath72 by fitting their approximate semi - numerical simulations of reionization . in ref.@xcite
, we showed by our analytical derivation and numerical lptr calculations that this fit underestimates the scale - dependent bias due to png significantly .
these differences are reflected in the smaller @xmath80 we obtain for lofar , ska , and omniscope .
( 2 ) the anticipated mwa512 configuration assumed by ref .
@xcite was also overly optimistic , while we adopt the current mwa128 configuration @xcite . to model the effect of png on the 21 cm power spectrum , ref .
@xcite assumed the simple functional dependence of ionized fraction on local overdensity , used for illustrative purposes by ref .
@xcite for the gaussian case , to derive an ionized fraction bias @xmath85 for png .
unfortunately , they incorrectly used @xmath85 to relate the 21 cm power spectrum to the matter power spectrum ( see ref .
this paper suggests two approaches to constrain @xmath0 with 21 cm power spectra from the eor .
if we take a conservative approach , i.e. assuming nothing about @xmath20 and @xmath86 , then the pheno - model can be employed for a single redshift bin to provide @xmath0 constraints with the same precision as a reionization model in which the reionization history is uniquely specified by a set of model parameters . however , using the esmr , we demonstrate that a well - motivated reionization model can improve @xmath80 in two ways : ( 1 ) a pathfinder measurement at a single redshift can best - fit the values of model parameters , which then can be used to estimate the desired redshift corresponding to @xmath87 , i.e. the `` sweet spot '' for cosmic - variance - dominated experiments .
this can help tune _ single - epoch _
observations for maximum precision .
( 2 ) _ multi - epoch _ measurements can be combined to improve @xmath80 .
we find that multi - frequency observation by ska can achieve @xmath1 , providing a new method to constrain png independent of cmb measurements , but with a precision comparable to _ planck _ s .
a cosmic - variance - limited array of this size like _ omniscope _ can achieve @xmath2 , improving current constraints by an order of magnitude .
these high precision observations may someday shed light on inflationary models .
we thank shahab joudaki and mario santos for additional information on their work in ref .
this work was supported by french state funds managed by the anr within the investissements davenir programme under reference no .
anr-11-idex-0004 - 02 , by u.s .
nsf grants no .
ast-0708176 and no .
ast-1009799 , and nasa grants no .
nnx07ah09 g and no .
nnx11ae09 g .
m. tegmark , and m. zaldarriaga , phys .
d * 79 * , 083530 ( 2009 ) ; m. tegmark , and m. zaldarriaga , phys .
d * 82 * , 103501 ( 2010 ) ; s. clesse , l. lopez - honorez , c. ringeval , h. tashiro , and m. h. g. tytgat , phys
. rev .
d * 86 * , 123506 ( 2012 ) | measuring the small primordial nongaussianity ( png ) predicted by cosmic inflation theories may help diagnose them .
the detectability of png by its imprint on the 21 cm power spectrum from the epoch of reionization is reassessed here in terms of @xmath0 , the local nonlinearity parameter .
we find that an optimum , multi - frequency observation by ska can achieve @xmath1 ( comparable to recent _ planck _ cmb limits ) , while a cosmic - variance - limited array of this size like _ omniscope _ can even detect @xmath2 .
this substantially revises the methods and results of previous work . |
important developments have been taking place in the growth of strained heterostructures based on si and ge@xmath0si@xmath1 layers @xcite .
the particular case of si layers pseudomorphically grown on relaxed ge@xmath0si@xmath2 buffers is rather interesting because of the large mobilities reported , exceeding the corresponding bulk @xmath3 values @xcite .
it has been shown that in this type of double heterostructures ( dhs ) the otherwise degenerate six @xmath4 valleys of bulk si are shifted in energy . if the dhs is grown along a high symmetry direction ( say , the @xmath5 direction ) , the two valleys along this direction are shifted downwards in energy , while the other four valleys are shifted upwards . as a result , we are led to two @xmath6 valleys with a bottom energy below the corresponding one for the bulk valleys and four @xmath7 valleys with a higher bottom energy .
the energy shift between the @xmath6 and @xmath8 valleys is empirically estimated as @xmath9 ev . in this kind of dhs , a quantum - well like band alignment in the si channel
was unexpectedly found in 1985@xcite , which is due to tensile strains in si and ensures high - mobility @xmath10 channel for the doped structure .
high mobilities are explained as a consequence of the low effective mass of the carriers in @xmath6 valleys and also because the scattering efficiency of the carriers ( by phonons and impurities ) is reduced .
the above mentioned facts are of great importance for device performance and this kind of structures seems to have a prominent future for applications in micro and optoelectronics @xcite .
monte carlo calculations of high electric field mobilities and drift velocities for electronic transport along the si channel of si / ge@xmath0si@xmath2 were reported by various authors @xcite and also applied in the study of modulation - doped field - effect transistor ( modfet ) structures @xcite .
more recently , hole transport parameters have been analyzed for this type of systems @xcite .
while the electronic transport properties of low field , low temperature si / gesi heterostructures are more or less well understood @xcite , the situation is not the same for high electric fields . for high field transport , the effects of size quantization are usually negligible and we can work within three dimensional models for the valley structure . in ref .
@xcite , size quantization was introduced in a 10 nm si channel , but the authors had to include up to six subbands in the @xmath11 valleys .
strain effects are introduced through the splitting energy @xmath12 between the @xmath6 and @xmath7 valleys , which ranges from @xmath13ev to @xmath14ev .
such relatively large values of the energy shift effectively reduce the intervalley phonon scatterings between the valleys ( in comparison with the unstrained si ) . in the present paper
we report results for high electric field electronic transport along the si channel of a si / gesi dhs .
we model the system in close analogy with the above mentioned works , taking into account the energy shift between the @xmath6 and @xmath7 valleys and neglecting size quantization . rather than monte carlo simulations
, we apply an analytical fokker - planck approach ( fpa ) , which treats transport as an energy - diffusion process in energy space .
this approach was proposed a long time ago @xcite as an alternative for the consideration of high - field transport in semiconductors .
the theory bears a semiclassical nature and has been recently revisited , the general formalism being discussed with further details @xcite . as a test of the fpa in the case of a well known semiconductor , the present authors applied the theory to bulk si , where both experimental data and monte carlo simulations for the high field drift velocity were successfully reproduced @xcite .
the current calculations can also be considered as a way of testing the fpa in a somewhat different system , which is of present day interest in high technology applications .
the fpa is applicable when the energy exchanges between the carriers and the surrounding medium ( crystal lattice + external field ) can be assumed quasicontinuous .
this latter condition apparently invalidates the method for highly inelastic scattering mechanisms ( as is the case of carrier scattering by optic and intervalley phonons ) .
however , if the carrier energy is large enough , the exchanged energy becomes certainly low compared with the energy of the carriers , and this is the case of high field transport .
hence , the fpa is assumed to be valid if the condition @xmath15 is fulfilled , where @xmath16 is the average carrier energy and @xmath17 the exchanged energy ( say , the phonon energy ) .
the fpa has the advantage of being analytical , and , whenever it can be applied , saves computational time and allows a more transparent physical interpretation .
of course , it can not compete in accuracy with monte carlo simulations .
however , we have found that the fpa leads to comparatively good results even in cases where several scattering mechanisms should be taken into account @xcite .
in contrast with monte carlo simulations , the analytical approach involves less realistic models for the analysis of a concrete semiconductor .
more details about the applied model and the fundamental theory are given in the next sections .
the paper is organized as follows . in sec .
ii we briefly summarize general theoretical aspects of our work , in sec .
iii details of the calculations for the si / gesi dhs are presented , in sec .
iv the results of our calculations are shown and comparison with previous works is made .
the fpa considers transport in the spirit of a drift - diffusion process in energy space . a certain distribution function ( df ) , @xmath18 , is defined which depends on the carrier energy @xmath19 and the time @xmath20 , such that @xmath21 gives the number of carriers at time @xmath20 having their energies in the interval @xmath22 $ ] , while the function @xmath23 represents the density of states ( dos ) .
the df obeys the equation @xcite : @xmath24 where @xmath25 .
\label{e2}\ ] ] in eq .
( [ e2 ] ) @xmath26 represents a certain `` drift velocity '' of the carriers in energy space and in fact gives the rate of energy exchange of the carriers with the surrounding medium , while @xmath27 is a kind of diffusion coefficient .
equation ( [ e1 ] ) has the form of a continuity equation for the carrier `` motion '' in energy space and leads to the conservation of the carrier `` flux '' . for a many valley system
it should be applied to each of the valleys separately and it is worth to note that eq .
( [ e1 ] ) does not describe intervalley couplings .
the quantity @xmath28 , given by eq .
( [ e2 ] ) , is thus the carrier current density in energy space . under steady - state conditions
the df is time independent .
moreover , we must have @xmath29 .
hence , we are led to @xmath30 = w(e)n(e)f(e ) .
\label{e3}\ ] ] we assume the existence of a phonon bath in equilibrium at the temperature @xmath31 .
the carriers interact with the phonons and the applied dc electric field @xmath32 .
we suppose , as an approximation , that the continuous exchange of phonons between the carriers and the phonon bath does not affect the thermal equilibrium of the latter .
the coefficients @xmath26 and @xmath27 are split as follows @xmath33 the label `` @xmath34 '' ( `` @xmath35 '' ) denotes the electric field ( phonon ) contribution to these coefficients .
we are obviously neglecting the intracollisional field effect .
the explicit form of these coefficients is obtained below . equation ( [ e3 ] ) has the simple solution @xmath36 \right\ } , \label{e5}\ ] ] where @xmath37 was neglected .
this approximation is very well fulfilled in all the cases of interest for us @xcite .
the practical usefulness of the fpa lies in the possibility of the _ analytical _ performance of the integral in eq .
( [ e5 ] ) .
let us apply the above formalism to electrons in the @xmath6 energy valleys of strained si of a si / ge@xmath0si@xmath1 dhs .
the @xmath6 valleys are ellipsoids of revolution in @xmath38-space with their revolution axis along the @xmath5 crystallographic direction , coincident with the growth direction of the dhs . to be specific ,
let us take @xmath39 , where the @xmath40 coordinates are taken along @xmath41 , @xmath42 and @xmath5 respectively .
the energy dispersion relation is given in the form @xmath43 . in order to take into account non - parabolicity of the band structure , we assume @xmath44 in eq .
( [ e6 ] ) @xmath45 ( @xmath46 ) is the transverse ( longitudinal ) effective mass of the bulk si conduction band electrons .
we suppose the masses are not changed by the strains in the si layer : @xmath47 and @xmath48 .
nonparabolicity is estimated also the same as in bulk si : @xmath49 ( ev)@xmath50 .
the herring - vogt transformation was applied in writing the right - hand side ( rhs ) of eq .
( [ e6 ] ) , where @xmath51 , @xmath52 and @xmath53 .
the explicit expressions for @xmath54 and @xmath55 can be directly taken from refs.@xcite .
hence , for @xmath54 we have @xmath56 where @xmath57 denotes the first derivative of the function ( defined in eq .
( [ e6 ] ) ) . for @xmath58 we shall consider one `` @xmath59 '' intervalley phonon responsible for transitions between the equivalent @xmath6 valleys .
then the relaxation time reads as @xmath60 @xmath61 , \label{e12}\ ] ] where @xmath62 is the step - function , @xmath63 ^{-1 } , \label{e13}\ ] ] and @xmath64 in eq .
( [ e14 ] ) @xmath65 is the semiconductor mass density , @xmath66 and @xmath67 are the phonon frequency and deformation - potential ( dp ) constant respectively for intervalley phonons of type @xmath68 . as a simplified model for our calculations
, we have considered just one @xmath68 phonon with an energy @xmath69ev and dp coupling constant @xmath70 ev / cm , which approximately corresponds to the average of the three @xmath68 phonons of bulk si reported in table vi of ref .
other bulk si parameters are also taken from the same reference . for ellipsoidal @xmath6-valleys of the si layer cb ,
the dos is given by @xmath71{% m_{t}^{2}m_{l}}$ ] . in calculating @xmath72
we use the expression @xmath73 , \label{e9}\ ] ] where `` _ abs _ '' ( `` _ em _ '' ) denotes phonon absorption ( emission ) by the electron
. we must remember that @xmath74 the `` _ abs _ '' ( `` _ em _ '' ) term in eq .
( [ e15 ] ) corresponds to the first ( second ) term at the rhs of eq .
( [ e12 ] ) for the case of intervalley `` @xmath59 '' phonons . hence , using eq .
( [ e9 ] ) and latter expressions for @xmath75 and @xmath76 , we can find the explicit form of @xmath77 .
intravalley optical phonons do not contribute to transition rates because the corresponding transitions are forbidden by the selection rules .
the contribution of intravalley acoustic phonons will be ignored in the present work . for high temperatures and high carrier energies they should provide a weak contribution to transport parameters and we shall limit ourselves to this case . as was remarked above , intervalley phonon scattering , strictly speaking ,
is beyond the scope of the fokker - planck equation in its standard form [ eq .
( [ e1 ] ) ] . after the absorption or the emission of an intervalley phonon
, the electron is dropped away from one valley into another one , and such abrupt transitions are not explicitly assumed in eq .
( [ e1 ] ) , where the electron number is conserved within each valley .
however , when we have equivalent valleys ( as is the case of @xmath6 valleys ) , we can still apply the fokker - planck equation in its standard form ; electrons dropped away from one valley fall into an equivalent one and vice versa , all happens as the electrons always remained within the given valley .
of course , if the valleys are not equivalent this argument no longer applies .
this is the case of electron transitions induced by the `` @xmath78 '' intervalley phonons between valleys @xmath6 and @xmath7 , which are not equivalent in the strained si layer of the si / gesi dhs . for such intervalley electronic transitions we must consider a different approach . in order to estimate the effect of electron intervalley transitions due to the interaction with ``
@xmath78 '' intervalley phonons we must realize that this effect is actually weak and should introduce just small changes into our final results . as for the case of `` @xmath59 '' intervalley phonons , we assume just one type of `` @xmath78 '' phonons with energy @xmath79ev and dp coupling constant @xmath80 ev / cm ( we again estimated these parameters as averages from the three corresponding `` @xmath78 '' phonons reported in table vi of ref .
@xcite ) . in all cases we considered zero - order intervalley phonons in our model
( a more realistic model must consider first order intervalley phonons in the way discussed in ref .
electron transitions due to the interaction with `` @xmath78 '' intervalley phonons from @xmath6 valleys up to @xmath7 valleys is formally switched on at @xmath81 and the df in each of the @xmath6 valleys evolves in time by the approximate law @xmath82 , where @xmath83 and the corresponding expressions for @xmath84 and @xmath85 are obtained from eqs.([e12 ] ) , ( [ e13 ] ) and ( [ e14 ] ) by means of the formal substitution : @xmath86 and @xmath87 .
after a time @xmath88 has elapsed , we obtain the result @xmath89 , \label{e20}\ ] ] where @xmath90 represents the df without taking the @xmath6-@xmath7 intervalley scattering into account .
of course , this is just a rough estimation of the effect and we are actually neglecting the inverse intervalley process ( from @xmath7 to @xmath6 valleys ) . due to the weakness of the effect
, we shall assume that this procedure is satisfactory .
for the parameter @xmath88 in eq .
( [ e20 ] ) , we make the reasonable estimate @xmath91 ^{1/2}. \label{e21}\ ] ]
from eq . ( [ e5 ] ) , with the explicit application of eqs .
( [ e11 ] ) and ( [ e9 ] ) , we are led to the df for @xmath6 valleys without the consideration of the @xmath6-@xmath7 intervalley scattering process .
the final result is @xmath92 where @xmath93 and @xmath94 are parameters ( dependent on @xmath95 and @xmath34 ) and @xmath96 is a polynomial in @xmath19 .
this structure is far from the maxwellian one .
the df describes a stationary non - equilibrium configuration where an electron temperature @xmath97 can not be defined . in eq .
( [ e22 ] ) we shall measure all energies in units of @xmath98 and @xmath99 for @xmath100 we obtain @xmath101 where the @xmath102 are coefficients explicitly dependent on @xmath103 . for @xmath104 the following result is obtained @xmath105 \nonumber \\
b_{2 } & = & \beta _ { t}\left [ ( n_{g}(t))^{2}2(8e_{0}^{2}+1)+(2ng(t)+1)(4e_{0}^{3}+8e_{0}^{2}+5e_{0}+1)% \right ] \nonumber \\ p_{2}(e , t ) & = & \sum_{i=0}^{4}a_{i2}e^{i } , \label{e25}\end{aligned}\ ] ] where the coefficients @xmath106 are now dependent on @xmath95 through @xmath107 .
all the coefficients @xmath108 are given in table i. in the interval @xmath104 we must distinguish three subintervals : @xmath109 , @xmath110 and @xmath111 , where @xmath112 .
in the latter two subintervals , the df shall be given by @xmath113 \quad , \quad j=3,4 , \label{e26}\ ] ] in correspondence with our approach to intervalley transitions between @xmath11 and @xmath7 valleys .
once we have determined the df , we can calculate the electron average energy by the expression @xmath114 in eq .
( [ e27 ] ) we just consider @xmath6 valleys , i.e. , we assume the @xmath7 valleys essentially deprived of carriers , an approximation that seems acceptable if the applied electric field is not extremely large . moreover , in eq .
( [ e27 ] ) we should be careful in partitioning the integration into the four energy subintervals mentioned above .
another important quantity is the drift velocity @xmath115 given by @xmath116 n(e)\right\ } de/\int \left\ { f(e)n(e)\right\ } de .
\label{e28}\ ] ] in eq .
( [ e28 ] ) we must partition the integration in the same way as in eq .
( [ e27 ] ) . we just considered the drift velocity from electrons in the @xmath6 valleys .
by direct application of the theory developed in the foregoing sections , we have made numerical calculations of the average electron energy @xmath16 and the electron drift velocity @xmath115 as functions of the temperature @xmath95 and dc electric field @xmath34 .
our results are essentially valid for high temperatures as far as intravalley acoustic phonon scattering was ignored . as it was remarked in sec .
i , the fpa is applicable for high electric fields , when the condition @xmath15 fulfils @xcite . in the examined si
/ gesi dhs , we take into account the relatively large energy shift between valleys @xmath6 and @xmath7 and consider that just the @xmath6 valley is substantially populated by carriers .
hence , intervalley transitions between valleys @xmath6-@xmath7 are assumed to be weak processes and treated within a relatively coarse approximation . as it was discussed in the previous sections , such transitions are actually beyond the scope of standard fpa .
all numerical parameters used in computations were taken from ref.@xcite and also shown in sec .
iii . in fig .
1 we show our results for the average electron energy @xmath16 as a function of electric field @xmath34 for @xmath117 k. two values of @xmath12 were considered : @xmath118 and @xmath119 ev . as expected , the results show a rather weak dependence on @xmath12 .
the involved carrier energies are large enough , thus ensuring the applicability of the fpa .
after comparison of fig . 1 with fig .
1(a ) of ref .
@xcite , we can see that a reasonable agreement was actually achieved .
it is important to notice that just one fitting parameter was applied in the @xmath120 of eq .
( [ e23 ] ) , which could be related to the overlapping integral describing the electron - phonon scattering probabilities .
this overlapping integral was assumed unity in the general formulas of sec .
ii , but actually it differs from unity when intervalley phonons are present . the linear behavior of @xmath16 for @xmath121 kv / cm should not be realistic because the fpa is not valid in the low - field regime as discussed exhaustively in previous works .
@xcite in fig . 2 we present the drift velocity @xmath115 as a function of the electric field for @xmath117 k. the continuous curve represents our calculations , while the dots were taken from fig .
2(a ) of ref .
we just considered the case @xmath122 ev . as a matter of fact , the curves for different values of @xmath12 are almost coincident ( in close agreement with the results shown in ref .
@xcite ) as can be expected from general physical grounds . for lower electric fields
our results deviate from those of ref .
@xcite , a reasonable result taking into account that the fpa is valid for high electric fields . however , for very high electric fields ( @xmath123 kv / cm ) we again notice an increasing deviation from the monte carlo results of ref .
the fpa is not able to describe the saturation value of @xmath115 in the very high electric field region . on the contrary ,
it is obtained a decreasing behavior of the drift velocity at a rate that becomes much higher as higher is the field . in our present treatment ( as well as in that of ref .
@xcite ) negative differential mobilities are out of question as far as the contributions of electrons from the @xmath7 valleys are not considered . as a final remark
, we should notice a very good agreement with the results of ref .
@xcite for a wide interval of electric fields . in fig .
3 we show the same plots as those of fig . 2 , but now for @xmath124 k. from a simple glance at fig .
3 , it is obvious that for such a low temperature the agreement between the fpa and the monte carlo results of ref .
@xcite ( fig .
2(b ) ) is much worse .
as it was discussed before , we have ignored intravalley acoustic phonons . even though this is not a limitation of the fpa itself ,
it is a difficult task to include them in the calculations .
furthermore , at low temperatures , quantum effects should become relevant and a quantum approach should be required .
hence , we stress that the fpa results are reliable just for high temperatures . in this work
we have compared our results with those from monte carlo simulations.@xcite other possible comparisons could be done .
for instance , we could compare with fig . 3 of ref.@xcite or ref.@xcite .
but all these results are in accordance with themselves and nothing essential would be added .
our results are also comparable to those of ref .
@xcite , where size quantization was examined . as it was said before ,
size quantization is of relevance when the carrier energies are low enough ( this is the case of low field transport as discussed in ref .
@xcite ) , but for the high carrier energies involved in high field transport , size quantization becomes irrelevant . in the revised literature we have not found available experimental data for this kind of system . as a general remark , we conclude that the fpa leads to results in acceptable agreement with those of monte carlo simulations in those intervals of temperature and electric fields where it is supposed to be valid .
this conclusion is true in spite of the more or less coarse simplifications we have to face within the limits of this method .
however , if wider intervals of electric fields or more accurate treatment is required , monte carlo simulations or some other equivalent numerical procedure should be necessary .
we acknowledge financial support from fundao de amparo pesquisa de so paulo ( fapesp ) .
f.c . is grateful to departamento de fsica , universidade federal de so carlos , for hospitality .
2 . drift velocity as a function of the electric field for @xmath117 k. our results are represented by the continuous while the dots were taken from monte carlo simulations of ref .
we have set @xmath122 ev . | we report calculations of high electric field transport for the case of a strained si / gesi double heterostructure ( dhs ) considering transport along the si channel and by applying the analytical fokker - planck approach ( fpa ) , where the process is modeled as drift - diffusion in energy space .
we limit ourselves to electronic transport in the conduction band of the strained si , where an energy shift between the otherwise degenerate six energy valleys characterizes the band alingment in the dhs .
intervalley phonon scatterings are considered while intravalley acoustic phonon scattering is ignored , leading to results valid for high enough temperatures .
our results are compared to previous theoretical works where monte carlo simulations were applied . a reasonable agreement between the two approaches
is obtained in the high electric field regime .
pacs : 72.10.di ; 72.20.ht ; 73.20.dx ; |
active galaxies can be classified in many ways , according to spectrum , luminosity , and morphology , and the result is a confused taxonomy that only partly reflects the underlying physics .
this situation has led to unification theories " in which various categories are linked by orientation , evolution or a gradient in physical characteristics .
( for reviews , see e.g. @xcite ) .
the one that concerns us in this paper connects type 1 ( broad - line ) and type 2 ( narrow - line ) objects by aspect . in this picture an active galactic nucleus ( agn ) contains a bright continuum source and a broad - line region ( blr ) surrounded by a dusty torus .
when viewed from near the pole the center is visible and a type 1 object is seen , whereas along the equator the view of the center is blocked by the torus , and a type 2 object is seen .
when the conditions are right the nuclear light ( continuum and broad lines ) escaping along the poles can be scattered and thus polarized , and seen by the observer .
miller and his collaborators ( @xcite ) , with the aid of spectropolarimetry , showed that many seyfert 2 galaxies do contain a hidden blr , and a detailed investigation of the seyfert 2 ngc 1068 ( @xcite ) showed that the scattered light is typical of a seyfert 1 nucleus .
thus unification by aspect is proven for many seyfert 1 and 2 galaxies . @xcite surveyed seyfert 2 s selected by fir fluxes and colors , which supposedly are isotropic quantities , and showed that at least 7 of 16 objects contain a hidden blr .
also , infrared observations that penetrate the dust and see the blr have shown directly that some type 2 seyferts have broad line regions ( @xcite ) . in this paper our interest centers on the powerful fr ii radio galaxies ( rg ) , and on radio - loud qso s ( quasars ) .
like seyferts , the rg come in types 1 and 2 ( blrg and nlrg , respectively ) , and there has been interest in seeing if they also can be unified by aspect .
they have the further virtue of containing a radio jet that presumably marks the dynamical axis of the black hole and accretion disk , providing additional geometric information .
quasars and rg are obviously anisotropic at radio wavelengths , and the viewing aspect has long been a major factor in their unification ( e.g .. @xcite ) .
the common deficit of observed ionizing photons relative to those seen by the line - emitting regions ( e.g. @xcite ) shows that rg also must be anisotropic at optical and uv wavelengths .
a statistical basis for unifying rg and quasars was provided by @xcite , who used the opaque torus scenario developed for seyferts .
barthel postulated that most fr ii rg and quasars are similar , and contain a continuum source and blr inside an opaque torus .
quasars are seen within about 45 degrees of the pole where there is a direct view into the center , and rg are seen at higher inclinations where the quasar is partially or entirely hidden .
blazars are seen when the line of sight ( los ) is close to the axis , and the rapid variability in flux and polarization due to variations in beamed synchrotron radiation can be seen .
many observations have now shown that much of this scenario must be correct .
broad h@xmath0 has been seen in the polarized light from the nlrg 3c 234 ( e.g. @xcite , hereafter t95 ) , 3c 321 ( @xcite ) , and cygnus a ( @xcite , hereafter o97 ) .
in all three objects an estimate of luminosity is consistent with their being off - axis quasars .
imaging polarimetry shows that in some rg the polarized regions have the azimuthal symmetry of a reflection nebula , as expected from an extended scattering region illuminated by a point source ( @xcite , o97 , @xcite ) .
also , as with seyferts , broad ir lines have been seen in a number of nlrg .
@xcite found broad p@xmath0 in 4 of 9 nlrg that they surveyed , indicating that some nlrg contain a blr hidden by dust , and would be called quasars if seen from a dust - free direction .
another aspect of unification is the alignment effect " , the tendency of galactic extended emission line regions ( eelr ) to lie along the radio axis ( see @xcite for a review ) .
at least part of the explanation involves an agn in which the radio axis , the dusty torus , and the photoionization cones are all roughly coaxial . this picture is generally confirmed with polarimetry , which shows that the reflection nebula is spatially coincident with the eelr .
jet - induced star formation might also contribute to the alignment effect ( e.g. @xcite ) . in 1994
we started a program of spectropolarimetric observations of powerful radio galaxies , to study unification and the alignment effect , and to study the geometry and kinematics of the agn .
we have already reported on 8 objects : pks 0116 + 082 , 3c 234 , fsc 10214 + 4724 , 3c 265 , 3c 277.2 , 3c 324 , 3c 343.1 , and cygnus a. ( @xcite , t95 , @xcite , @xcite , o97 ) in this paper we give the combined results for the 13 objects for which @xmath1 ; i.e .. , those in which h@xmath0 is available , and discuss the polarimetry , unification , geometric effects , and the continuum radiation . in general we assume that the scattering is due to dust , which undoubtedly is present , at least near the nucleus , because of the observed extinction .
the relative importance of electron and dust scattering is undetermined , since dust can mimic electrons in producing wavelength - independent scattering in the optical band .
the organization of this paper is as follows . in
2 we describe our sample of fr ii galaxies , 3 gives some observational details , and 4 and 5 give a general view of the results .
section 6 presents a model for the continuum radiation for blrg .
details of individual objects are in 7 and 8 . in 9
we discuss the rotations in @xmath2 ( polarization position angle ) observed in 3c 227 and 3c 445 .
section 10 contains further discussions of direct vs scattered light , unification , and radio - optical connections .
we choose @xmath3 and @xmath4 for all calculations .
we shall make use of both the full stokes parameters @xmath5 and the normalized parameters @xmath6 ; the use of upper and lower case letters should minimize any confusion .
we have been measuring the polarimetric properties of radio galaxies with the 10-m w.m .
keck telescopes , and define those in which h@xmath0 is visible as the low - z sample " . to this keck group we add two objects observed with the 5-m hale telescope at palomar observatory .
the total sample , listed in table 1 , comprises 13 fr ii rg with @xmath1 5 blrg and 8 nlrg .
3c 234 is often called a blrg because broad h@xmath0 is readily seen in f@xmath7 , but we list it with the nlrg because the broad line is predominantly composed of scattered light ( see 4.1 ) .
column 6 of table 1 gives an estimate of an upper limit to the interstellar polarization ( isp ) towards each source .
this value is obtained by estimating the extinction towards each object from the maps prepared by @xcite , and using the empirical formula @xmath8 ( @xcite ) .
cygnus a is at low latitude and is discussed by @xcite ; see also o97 .
the isp effects are small in this case .
possible isp effects in 3c 33 , 3c 195 , and fsc 2217 + 295 are discussed in 8 and 9 .
references to previous polarimetric work are in column 7 of table 1 .
our sample is not an approximation to a flux - limited or volume - limited sample , as most of the objects were chosen because they were known to be polarized .
3c 105 and 3c 135 were observed because they are fr ii rg that fitted into a slot in the observing schedule .
3c 357 was observed because its hst image ( @xcite ) shows an off - nuclear emission region that we thought might be polarimetrically interesting .
we shall make use of work by h96 who observed p@xmath0 in a complete sample of 11 rg ( see 10.2 ) , 3 of which are also in our sample .
this will give our results more generality than they would have by themselves .
the keck observations were made with the polarimetry module ( @xcite ) in the low - resolution imaging spectroscope ( @xcite ) .
the polarimeter contains a rotatable half - wave plate and a calcite beamsplitter .
the stokes parameters are measured by taking 4 exposures at waveplate position angles ( pa ) of @xmath9 , and combining the spectra with the standard self - calibration procedure ( @xcite ) , using the computer reduction system vista . at each epoch
we observed standard polarized stars and standard null stars as a check on the system and to establish the reference for pa determinations .
the standard stars are taken from the lists of @xcite or have been referenced back to that list by our own measurements with the keck polarimeter .
the relative accuracy in pa determinations is about 0.5 .
the pa calibration is discussed in detail by @xcite .
for spectroscopy we used a grating with 300 g / mm , blazed at 5000 .
this gives a dispersion of @xmath10 per pixel and a resolution of @xmath11 .
the wavelength coverage was typically 3900 to 8900 .
the slit was @xmath12 wide , except where noted in table 2 .
we did not observe with an order - blocking filter .
in most cases the spectra are flat or rise to the red , and errors due to second - order light are negligible .
we did observe the flux standards with and without the order - blocking filter , as most of them are hot white dwarfs and the flux calibration would otherwise have been in error . for imaging we replaced the grating by a mirror and inserted a b - band filter into the beam .
the scale is @xmath13 per pixel .
the imaging polarimetry procedure is the same as for spectroscopy , with the simplification that there is no spectral extraction .
when necessary for either spectroscopy or imaging we improved the s / n by binning the fluxes rather than by averaging the stokes parameters themselves , which can lead to bias ( @xcite ) .
we typically took 15-minute exposures so that a full 4-exposure set took an hour plus readout time .
the field of view of the keck polarimeter is about @xmath14 .
this limits the useful length of the slit for spectroscopy , and for all observations it can make sky subtraction difficult , as some galaxies nearly fill the field . in such a case the apparent polarization @xmath15 may be higher than the value a larger field would have given ; this is analogous to increasing @xmath15 for a nuclear source by narrowing the slit and allowing less diluting starlight into the the measurement .
at palomar we used the double spectrograph ( @xcite ) to do spectropolarimetry , but not imaging polarimetry , on the 5-m hale telescope .
the polarimeter is similar to that used at keck ( @xcite ) . in the red camera
the grating has 158 g / mm and dispersion approximately 6 per pixel , and in the blue camera the grating has 316 g / mm giving approximately 2.1 per pixel .
calibration and data reductions were done in a fashion similar to those used at keck .
table 2 gives the log of all the observations .
the imaging observations on 1995 december 16 were compromised by poor seeing that ranged up to @xmath16 .
we reobserved 3c 195 and 3c 234 on 1996 april 16 and only the later data are included in this paper .
four rg , pks 0116 + 082 , 3c 109 , 3c 227 and fsc 2217 + 259 , were observed only in december 1995 , and those data are included here except for pks 0116 + 082 , which was presented in cohen et al .
the high and variable seeing compromises the angular resolution , but does not further affect the polarization observations .
we display the combined spectropolarimetry results in figures 1 and 2 , where the objects are arrayed , roughly , in order of decreasing visibility of the broad h@xmath0 line in f@xmath7 . in figure 1
the left panels show the total flux f@xmath7 and the right panels show the polarized flux " defined as @xmath17 .
this quantity is sometimes called the stokes flux " , and in simple cases it is proportional to the scattered flux .
the spectra of f@xmath7 and @xmath18 are shown on a logarithmic scale .
note that corresponding plots in figure 1 have the same logarithmic range , so that the relative enhancement of h@xmath0 can be more readily assessed .
figure 2a shows the fractional linear polarization @xmath15 and figure 2b shows the polarization position angle @xmath2 .
when the s / n is high ( @xmath19 ) we can estimate @xmath15 as @xmath20 where @xmath21 and @xmath22 are the normalized stokes parameters and @xmath23 is the error in @xmath21 or @xmath22 calculated from photon statistics and the normal propagation of errors ( @xcite ) . in most cases
, however , s / n is not that high , especially at the blue end of the spectrum . instead of attempting to debias the square - root , we prefer to rotate @xmath21 and @xmath22 by @xmath24 , where @xmath25 is a smooth approximation to the measured @xmath2 ( @xcite ) .
the resulting @xmath26 is a useful estimate of @xmath15 .
hereafter we use the symbol @xmath15 for all estimates of the polarization . in figures 1 and 2
we show spectra for 3c 109 , 3c 234 and cyg a that have been published before ( @xcite , hereafter g92 ; t95 , o97 ) and are repeated here to give a more complete impression of the range of possible behaviors .
3c 105 had no measurable polarization and spectra for it are shown in 8.3 .
the polarization of 3c 357 is particularly noisy and it is omitted from figures 1 and 2 but is shown in figure 3 and discussed in 8.7 .
generally the spectra are for a region centered on the nucleus , except for 3c 33 where the extraction window was centered on a strongly - polarized region approximately 3 north - east of the nucleus , and 3c 357 where the window is approximately 1.5 west of the nucleus .
the length of the window is given in table 3 , which gives some results from the polarimetry .
the averaging interval , 5400 5600 , contains no strong emission lines , so the values of @xmath15 and @xmath2 represent , approximately , the v - band continuum polarization .
however , we show in 5.2 that many of the nlrg contain reflection nebulae , in which @xmath2 varies with position . in these objects
the measured @xmath2 may depend on the location of the slit and the extraction window . for most objects an elliptical galaxy template was fit to the absorption features and subtracted from f@xmath7 to get the corrected flux " f@xmath27 .
the corrected polarization @xmath28 is in table 3 .
we used ngc 6702 as the template galaxy except for 3c 234 and cyg a where ngc 821 was used .
two blrg , 3c 382 and 3c 445 , have so little starlight that we were unable to determine an accurate galaxy component other than @xmath29 .
we were unable to fit 3c 109 properly , but the absorption lines are weak and the galaxy fraction must be small .
figure 3 shows curves of f@xmath30 and @xmath31 for 7 objects ; those for 3c 234 and cygnus a have already been published ( t95 , o97 ) . at this point
we discuss the main features of figures 1 - 3 .
details will be presented in later sections .
it has been customary to define a broad - line radio galaxy merely as one that has prominent broad lines , e.g. the balmer series , that are substantially broader than the forbidden lines e.g. [ oiii ] .
this simple definition , however , takes no account of equivalent widths , and is subject to the vagaries of observation and signal - to - noise ratio .
further , it does not allow for differing geometric or physical circumstances . in figure 1a , 3c 234 shows prominent broad h@xmath0 and easily seen broad h@xmath32 .
these lines were noted long ago and thus 3c 234 is usually called a blrg ( @xcite ) .
note , however , that the broad lines are noticeably weaker than those in the top 5 galaxies in figure 1a , and 3c 234 has also been called an nlrg .
@xcite argue by analogy with the type 2 seyfert ngc 1068 that the broad lines are polarized like the continuum and are predominantly due to scattered light , and that 3c 234 should be called a type 2 object ( see also t95 ) .
we agree that the broad lines are mainly due to scattering , because broad h@xmath0 and h@xmath32 are more prominent in @xmath33 than in f@xmath7 .
thus we discuss 3c 234 with the nlrg . however , it is intermediate between type 1 and type 2 objects , and has other differences from the blrg and nlrg in addition to the intermediate equivalent width of the broad lines .
we discuss this in 8.5 . in figure 1a the blrg
have been arrayed in order of decreasing visibility of broad h@xmath0 , but it can be seen that the order is also one of increasing redness and @xmath15 .
3c 382 has a blue quasar spectrum and is weakly polarized ; fsc 2217 + 259 is a red iras galaxy and is highly polarized .
the correlations are shown in figure 7 , along with calculations made with a simple model which is presented in 6 . briefly , the model assumes that the blrg continuum ( after starlight subtraction ) is a combination of two components , a nearly unpolarized @xmath34 ray that has a view of the nuclear continuum source and blr but has extinction and reddening due to dust , and a highly polarized @xmath35 ray that has less reddening .
the net polarization increases as the ratio of scattered to direct radiation increases , so the direct ray bluens the polarization and reddens the total flux .
this model is similar to the one used by @xcite for iras 13349 + 2438 .
application to individual objects is given in 7 .
the continuum model can not be used for the nlrg because they have an additional important component of radiation .
most of the nlrg , along with many seyfert 2 galaxies , have broad lines that are more highly polarized than the neighboring continuum , even after the subtraction of starlight ( figure 3b ) .
this led tran ( 1995 ) to consider two featureless continuum components ; the first , fc1 , is due to scattering and is polarized , while the other , fc2 , is assumed to be unpolarized .
fc2 differentially dilutes the continuum and h@xmath0 ; this leaves @xmath36 .
the relative levels of fc1 and fc2 can be estimated by requiring the broad h@xmath0 emission line to have the same intrinsic polarization as the neighboring fc1 .
this is useful if it can plausibly be maintained that the agn light ( continuum and broad lines ) is all polarized at the same level , by scattering .
we think that this is a conservative assumption , as it is unlikely that the broad lines could be intrinsically more polarized than the nuclear continuum .
they surely come from a larger region ( @xcite ) , and are more likely to be intrinsically polarized less than the continuum , as in 3c 382 ( see also @xcite ) . in 3c 234
the decomposition into fc1 and fc2 ( t95 ) showed that the continuum may consist of three components in addition to unpolarized starlight : balmer nebular continuum radiation from the nlr , nuclear light in which the continuum and broad lines are polarized the same ( fc1 ) , and an fc2 whose origin is unclear .
the reliability of this model was enhanced when it turned out that broad h@xmath32 had the same @xmath15 as its neighboring fc1 , even though h@xmath32 was not used in the construction .
cygnus a yields a similar result ( o97 ) , and in this case fc2 can be interpreted as diluting radiation from hot young stars ( @xcite ; see also @xcite ) .
nearly every object has @xmath15 rising to the blue ( figure 2a ) . for most objects
this is mainly due to the reduction of dilution of @xmath15 by unpolarized light from an old stellar population .
( this point has been made numerous times ; see e.g. t95 , and @xcite ) .
it is generally said that most low - z rg have low @xmath15 whereas most high - z rg have high @xmath15 , and that this is due to the different rest - wavelength ranges covered at low and high z. figure 2 shows this wavelength dependence : at 5500 all but 3c 109 , fsc 2217 + 259 and 3c234 have @xmath37 , whereas below 4000 all but 3c 382 , 3c 445 and 3c 135 have @xmath38 .
however , figure 3b shows that @xmath31 still rises to the blue in most objects .
as remarked above , this could be caused by another diluting component ; e.g. a reddened direct ray or fc2 .
the rise of @xmath31 to the blue , especially in the uv , means that the polarized flux @xmath33 has a steeper spectrum than the galaxy - corrected flux f@xmath30 . in @xmath39 3c 382 and 3c
321 both have an index @xmath40 ( f@xmath41 ) , at least blueward of h@xmath0 .
this is part of the evidence that 3c 321 , an nlrg , has a quasar - like central continuum source .
see 8.6 .
the position angle @xmath2 ( figure 2b ) is remarkably constant in many of the objects . in fsc 2217 + 259 , for example ,
a linear fit to @xmath2 gives a slope @xmath42 which is at most @xmath43 across the spectrum . in 3c 234
a similar fit gives a change @xmath44 across the spectrum .
in these objects there can be little polarization by intervening dust extinction , either in our galaxy or the host galaxy , because @xmath15 itself changes strongly with wavelength and the observed @xmath2 would rotate noticeably , unless there were a fortuitous alignment of position angles . a few objects do have a significant rotation in the position angle of the continuum .
the two objects with double nuclei have large rotations ; 3c 321 rotates @xmath45 between 3550 and 7000 , and cygnus a has @xmath46 .
the cygnus a rotation is explained with three components of continuum radiation , with different @xmath15 s and @xmath2 s , that combine to give the observed values ( o97 ) . in all the polarized nlrg except 3c 357 , @xmath15 rises in broad h@xmath0 ( figure 2a ) .
this is due to differential dilution by starlight and other components .
it is this increase in @xmath15 that makes broad h@xmath0 more prominent in the polarized flux @xmath47 than in @xmath48 ( figure 1 ) .
3c 357w is the only polarized object that does not show broad lines .
however , this object is faint and the telluric a band falls on the red wing of h@xmath0 , so this result is inconclusive ( see 8.7 ) .
3c 382 is different from the other blrg as it has a quasar spectrum , with a weakly polarized continuum and little or no starlight or narrow lines .
broad h@xmath0 and h@xmath32 are weak in @xmath49 , and are nearly unpolarized .
weak polarization of the broad lines relative to the continuum is typical of quasars ( @xcite ) .
the sequence in figure 1b shows that broad h@xmath0 is similar in polarized flux from 3c 227 to at least 3c 33 . in these cases
the polarized flux is giving us a look deep into the interior of the object , by scattering .
we also see the interior of the blrg directly , but with extinction ( see 6 ) .
this may be true also for 3c 234 ( t95 ) . in 3c 234 , and in some of the blrg , the los is probably near the edge of the obscuring torus .
we suspect that the torus does not have a sharp boundary , and that there is a range of inclinations where the interior can be seen through dust . in general
the narrow lines are only weakly polarized .
this is readily seen in figure 2a , where many objects show strong reductions in @xmath15 at the locations of the narrow lines .
this is most striking in 3c 234 where more than a dozen narrow lines show the effect strongly .
narrow h@xmath0 shows up prominently as a reduction in @xmath15 in 3c 234 , 3c 33ne , and 3c 195 because these are the objects that are dominated by narrow lines in total flux but by broad lines in polarized flux .
all the objects except 3c 109 , 3c 234 , and 3c 382 show narrow emission lines in @xmath49 ( figure 1b ) , so these lines are polarized but at a low level .
the possible polarization of the [ o iii ] lines in 3c 109 is discussed in g92 .
weakly polarized [ o iii]@xmath50 is seen in most objects , several objects show polarized narrow h@xmath0 , and cyg a shows several other species .
[ o ii]@xmath51 is not reliably identified in any of the @xmath49 spectra , and is essentially unpolarized
. the origin of [ o ii ] must be at or outside the scattering region , whereas the other narrow lines arise from deeper locations and are partly seen by scattering .
the possible anisotropy of [ o ii ] and [ o iii ] and its interpretation in terms of unified agn models has been discussed by e.g. @xcite , and @xcite .
we took polarimetric images of 11 objects ( table 2 ) .
earlier observations of this type are referenced in table 1 .
our current work shows that the polarization images fall into two main classes , those with compact and those with extended polarized regions .
there is a strong correlation between this image class and the spectral classification : the blrg are compact and the nlrg are extended . the compact systems are shown in figure 4 .
the innermost contour is 50% of the peak for every object , and except for fsc 2217 + 259 the next contour is 5% of the peak ( see the caption ) .
all the objects except fsc 2217 + 259 have @xmath52 , consistent with seeing .
3c 109 and 3c 227 have been called n " galaxies ( @xcite ) , indicating that they contain a brilliant point - like nucleus .
other n galaxies in our sample include 3c 135 and 3c 234 , shown in figures 5 and 6 , and 3c 445 , for which we have no imaging data .
3c 382 has a strong nucleus although apparently it was not called an n system . in
7.4 we give evidence that it was exceptionally bright at our epoch .
3c 382 , 3c 109 , and 3c 227 are closely similar .
the galaxies are dominated by a bright point source at the nucleus , and the polarization patterns are simple .
the vectors are roughly parallel and decrease radially by a small amount a factor of 0.8 in 3c 109 and 3c 382 , and 0.6 in 3c 227 .
these patterns can be explained with a polarized point source broadened by seeing and diluted by galaxian starlight and by spatial averaging .
there is no need to invoke scattering by an extended distribution of dust or electrons to explain the observations .
fsc 2217 + 259 is highly polarized and consequently has high s / n .
the vectors are accurately parallel and decrease radially by a factor of two or three .
it is similar to the other blrg and the polarization pattern could again be due to seeing plus dilution , although the angular extent is greater than for the others and scattering in an extended medium may also play a role .
this object is very red and we are seeing the nucleus through dust , as discussed in 7.4 . in 3c 382 @xmath53 and the scattering region is largely interior to the blr .
it may not be coincidental that 3c 382 has the weakest polarization of the blrg . in 3c 109 and fsc 2217 + 259 @xmath54 and @xmath2 does not change across h@xmath0 ; the main scattering region must be well outside the blr . in 3c 227 and 3c
445 @xmath2 rotates in h@xmath0 ; this is discussed in 9 .
images for the nlrg are shown in figures 5 and 6 , except for cygnus a , which has been extensively discussed elsewhere ( o97 ) . in figure 5
the top panels show the total flux , the polarization vectors , and the radio axis .
the lower panels show contours of polarized flux @xmath17 , with the slit location .
figure 6 shows total flux and polarization vectors for the two nlrg that have weak polarization and do not have extended polarization regions . in 3c 33 , 3c 195 ,
3c 321 , and cyg a ( o97 ) the polarized regions are in fans " on opposite sides of the nucleus , and the polarization vectors are roughly azimuthal ; i.e. they are tangential to circles around the center .
this is the sign of a bipolar reflection nebula .
a central source illuminates material in a bicone , and the scattered light is polarized perpendicular to the plane of scattering .
although the perpendiculars to the polarization vectors accurately define a center when averaged , seeing plus spatial averaging tilt the individual vectors away from perpendicular . in all the objects with polarization fans
the polarization is weak at the nucleus , and has a strong radial gradient .
the sign of the gradient is opposite to that for the blrg , where @xmath15 decreases with radius .
the gradient must be partly due to dilution by unpolarized starlight , which is more strongly concentrated to the nucleus than is the scattering material .
however , the effect is also seen in 3c 265 with z=0.811 ( @xcite ) . in this case
the polarization image was measured in the rest - frame uv , and dilution by an old stellar population would have had little importance . seeing and spatial averaging
also affect the pattern , but they generally act to reduce the gradient .
three of the four objects in figure 5 and also cyg a ( o97 ) ; i.e. all the strongly polarized nlrg except for 3c 234 , have separated peaks in @xmath17 . since starlight is unpolarized it has no effect on @xmath17 and we must look elsewhere for the cause .
we suggest it is due to the presumed geometry , an illuminated cone whose apex is shielded by an opaque torus . near the apex
the resolution element is larger than the cone and the vector sum of the polarizations is reduced because they have a variety of angles , all perpendicular to the radius .
the central decrease could also be caused by obscuration of the polarized light .
3c 234 is special .
the polarization pattern is extended but does not show fans , and @xmath15 has a steady ew gradient with a peak east of the flux peak .
@xmath17 has a maximum at the center .
this pattern is different from both the blrg and the nlrg .
the vectors show a tendency for circular symmetry , although not as strongly as the other objects , and the pattern may be due to scattering of nuclear light .
this is discussed further in 8.5 .
in this section we first discuss the spectral index and balmer decrement for the blrg in terms of one reddening screen , and then present a two - component model that also attempts to explain the polarization . in 7 the results are applied to the individual sources .
assume that there is a dust screen in front of the blrg which produces the same extinction in the continuum and the lines .
then we have two measures of reddening : the ratio of broad h@xmath0 to broad h@xmath32 , and the ratio of the continuum fluxes at h@xmath0 and h@xmath32 .
the data plotted in figure 7a show that these quantities are correlated .
they have been corrected for galactic reddening , and 3c 227 , 3c 234 , and fsc 2217 + 259 have been corrected for an old stellar population ( the other objects have @xmath55 ) . the continuum spectral index @xmath0 between h@xmath0 and h@xmath32 is shown in the top scale . in
fig 7 the order of the objects ( left to right ) is the same as the vertical order in figure 1 except that galactic dereddening has moved 3c 109 to the left of 3c 445 .
baker ( 1997 , fig .
16 ) shows a similar diagram for radio - loud quasars .
the range of balmer decrement in figure 7a is about the same as in baker , but fsc 2217 + 259 has a considerably redder spectrum than any of her quasars .
four of the five blrg ( excluding 3c 109 ) form a sequence that is parametric in reddening .
we model this relationship by assuming that the intrinsic balmer decrement of the broad - line clouds is 3.0 ( case b ) .
the reddening line for this model depends on the intrinsic spectral index @xmath56 of the continuum source , and in figure 7a we plot solid lines for @xmath57 .
the great majority of quasar indices fall within this range ( @xcite ) .
the dotted lines show extinction @xmath58 in magnitudes , calculated with the @xcite extinction law with @xmath59 . the model can be adapted to all the blrg except 3c 109 .
however , radio loud quasars ( excluding css sources ) appear to have dereddened intrinsic indices close to 0.5 ( @xcite ) , and fsc 2217 + 259 lies rather far from that value
. it may be that it has an exceptional index , but it may also be that the model is too simple in assuming that the extinction is the same for both the continuum and broad lines .
if the extinction to the blr were more than to the continuum source , then fsc 2217 + 259 could be accommodated with an index near 0.5 .
indeed , the extra extinction could be associated with the blr itself . for
a discussion of dust in the blr see e.g. @xcite .
another alternative is that the intrinsic balmer decrement is substantially higher than 3.0 @xcite .
3c 109 is well outside the range of the model .
it is exceptional , either in the intrinsic balmer decrement or the intrinsic index ( or both ) .
alternatively , the model may not apply to 3c 109 , at least not in this simple form .
3c 234 is shown for comparison only ; the model can not be used for the nlrg because the scattered light arises far from the nucleus and can not be presumed to be behind the nuclear screen .
also , the nlrg show an important radiation component , fc2 , that is not accounted for here .
figure 7b shows the galaxy - corrected continuum polarization at 5500 ( rest ) vs @xmath60 for the five blrg and 3c 234 .
there is a trend for the polarization to increase with reddening .
@xcite made a similar plot for 13 blrg ; their sample includes all of our objects except fsc 2217 + 259 .
they explained the correlation between @xmath15 and the balmer decrement as either dichroic absorption or scattering with extinction . the hidden nucleus model for nlrg leads us to consider a more elaborate model to explain the correlation .
it consists of two components : an unpolarized direct ray , and a ray from the nucleus that is scattered and highly polarized .
the components both have extinction , but the direct ray has more .
this is consistent with the presumed blrg geometry in which we are near the edge of the cone .
the direct ray must penetrate more dust than the scattered ray that passes closer to the axis .
the effect of the model is to redden the continuum , increase the balmer decrement , and bluen the polarization by differential dilution .
the model is similar to that used by wills et al .
( 1992 ; see also @xcite ) , except we assume that the extinction of the scattered component arises from dust inside the scattering volume itself .
this has the advantage that the scattering and extinction are coupled , and there is one less free parameter .
the direct ray can be expressed as @xmath61 , where @xmath62 is the unobstructed flux from the nucleus , and @xmath63 is the optical depth . to properly describe the scattered ray
we should specify a geometry and integrate through the scattering volume .
such calculations have been done by @xcite for a bicone uniformly filled with dust , but we merely make the simple approximation used by fosbury et al .
we assume that the scattered light can be expressed ( approximately ) as @xmath64 , where @xmath65 is a geometric factor and @xmath66 is a scattering efficiency that depends on the size distribution of the dust particles and on their composition .
each ray passes through a dusty region before and after it is scattered , and the factor @xmath67 accounts for the extinction in the scattering volume .
the optical depth @xmath68 is proportional to an extinction efficiency @xmath69 that is roughly proportional to the scattering efficiency @xmath70 , provided the wavelength is not near the graphite resonance at 2200 .
the net result of these assumptions is that the scattered ray can be expressed as @xmath71 the function @xmath72 has a maximum at @xmath73 , and most of the contributions to f@xmath74 come from regions near @xmath75 . for simplicity
we now set @xmath76 and eq ( 1 ) becomes @xmath77 where @xmath78 is independent of wavelength except for geometric factors associated with the change of the @xmath79 surface with wavelength .
@xmath78 is an adjustable parameter ; we expect @xmath80 and in the models shown below we use @xmath81 .
the scattered flux @xmath82 can be highly polarized ; the curves by @xcite show that the intrinsic polarization on @xmath82 , @xmath83 , can be as high as 50% .
the apparent polarization @xmath84 is @xmath85 equation ( 2 ) states that the scattered flux is proportional to the incident flux ; what we have done is to crudely justify the common assumption that the intrinsic polarization is independent of wavelength . the calculations by manzini & di serego alighieri
, however , show that @xmath86 generally rises to the red .
the model can be parameterized in various ways . in figure
7b we show lines of constant @xmath87 ( solid ) , and horizontal lines of constant extinction for the direct ray , @xmath88 ( dotted ) , together with the corrected data . in this example
the fixed parameters are @xmath89 and @xmath90 .
the model also predicts the balmer decrement and this is shown in figure 7c , which is analogous to figure 7a .
the balmer decrement and the index are limited because @xmath91 , and as @xmath88 increases beyond 3 mag the scattered ray becomes dominant .
3c 109 and fsc 2217 + 259 are outside the range of the model in figure 7c , and they are at the extreme values of @xmath87 in figure 7b .
the blrg of course must have a range in their intrinsic properties , but it is difficult to fit 3c 109 and fsc 2217 + 259 into the scheme described here . enlarging the model by adding excess extinction to or in the blr is one way to accommodate them .
another is by postulating the existence of an additional blue continuum component . in both cases
it must be done in a way that preserves the near equality of @xmath31 for h@xmath0 and the continuum ; i.e. the equivalent widths of the broad lines must be similar for all the rays . in studying the reddening of radio - loud quasars , @xcite found analogous discrepancies .
she suggests that excess reddening to the blr could be the explanation , but that other factors could also be responsible .
optical depth effects that increase the intrinsic balmer decrement could be part of the answer , as could dichroic dust extinction .
see 7.1 .
the point to this exercise is to show that the simple two - component model is able to explain qualitatively the observed spectra and polarimetry of the blrg .
further insight could be obtained with detailed calculations and a specific geometric model , and by examining the full spectrum rather than just the two - point spectral index .
the palomar spectropolarimetry of 3c 109 is discussed by g92 who interpreted the polarization as due to transmission through aligned dust in the host galaxy . in view of the success of the two - component model
, we now think that the origin of the polarization in 3c 109 should be reinvestigated .
3c 109 shows a serkowski - type curve of @xmath92 ( figure 2a ) , but the data are noisy below 5000 .
the appropriate way to discriminate between scattering and dichroic dust extinction is by extending the spectropolarimetry into the uv . in figures
7a and 7c 3c 109 stands apart from the other blrg .
we can describe this by saying that the balmer decrement is higher than expected from the continuum slope .
this could result from an intrinsically high balmer decrement , or from extra reddening to the blr .
alternatively , we could say that the continuum slope is bluer than expected from the balmer decrement , and this could result from an extra blue continuum component that is not in the model .
evidently there is a number of possibilities for explaining the discrepancy ; and , in addition , there remains the possibility that dichroic extinction plays an role .
@xcite have done extensive modeling of the eelr in 3c 227 .
they show that the velocity field can be modeled ( roughly ) with a bicone of opening angle @xmath93 , tilted @xmath94 to the plane of the sky , and rotating about an axis offset @xmath95 from the cone axis .
the projected bicone axis is at @xmath96 and the projected rotation axis is at @xmath97 .
we look inside a cone but near the edge .
their analysis of the emission lines shows that there is a photon deficit , and they suggest that 3c 227 harbors a quasar .
although the broad lines are prominent the quasar is hidden " because most of the ionizing radiation is not directly visible . the optical continuum radiation is polarized at @xmath98 , which is parallel to the bicone axis .
this is perpendicular to the orientation expected from scattering in the cone , but is consistent with the orientation expected from an optically thin disk coaxial with the cone .
the radio axis is at @xmath99 , which does not seem to be closely related to any of the above angles .
the broad lines are polarized at @xmath100 ( see 9 ) which is @xmath95 ( in projection ) from the cone and continuum radiation axes .
@xmath2 shows notable rotations in the broad balmer lines ; we discuss them in 9 .
the rotation in h@xmath0 has already been noted by @xcite . the spectrum of 3c 227 can be qualitatively understood with the one - component model in 6 . from figure 7a
, the balmer decrement implies 1.1 magnitudes of extinction , and the color is then explained with @xmath101 , the typical value for radio - loud quasars . to also explain the polarization , especially the fact that @xmath31 rises to the blue , we need to use the two - component model .
figure 7b shows that @xmath102 .
these particular values can be modified by varying the other parameters in the model , but @xmath103 remains somewhat low . with @xmath104 , @xmath105 and @xmath89 the calculated apparent
polarization rises from 1.7% at 7000 to 4.6% at 4000 .
this rise is steeper than shown in figure 3b , but the result is qualitatively correct . to use the model to match the polarization , @xmath58 would have to be reduced to near 1 mag , and the product @xmath106 would have to be doubled .
the point - source polarization image in figure 4 supports the scenario of @xcite described above , in which we are looking into the cone but near the edge .
their photon deficit is ... about a factor 2 for regions closer than 50 kpc , and a factor 13 for those located at much larger distances . "
our estimate of 1 2 mag for the v - band extinction to the nucleus is in rough agreement with their estimate of the photon deficit .
they conclude that the intrinsic luminosity of 3c 227 could put it in the quasar class .
our imaging observations of 3c 382 were saturated in three of the four exposures in the @xmath107 beam of the polarimeter , even though we reduced the exposure substantially from the expected value .
they were not saturated in the @xmath108 beam because the throughputs of the two sides differ by about 5% . and ,
as seen in figure 4 , the image shows the hexagonal diffraction pattern due to the secondary support and the mirror shape .
clearly , 3c 382 contained an unusually bright point - like nucleus during our observations .
that night we also observed the standard stars pg 1525 - 071a and b ( @xcite ) and used them as references .
the 2 stars were self - consistent to better than 0.01 mag .
we measured fluxes within a @xmath109 aperture .
the stars were observed about 4 hours before 3c 382 .
two other radio galaxies were observed between the stars and 3c 382 , and each had 4 exposures that show variations of 0.05 mag .
the night , however , was not photometric . with this caveat
we find @xmath110 and @xmath111 ( allowing 0.3 mag for galactic extinction ) , for the night of 16 april 1996 ( ut ) .
3c 382 shows double peaks in h@xmath0 ; this has been interpreted in terms of a rotating disk ( @xcite ) . we made the spectropolarimetric observations on 19 may 1996 ( ut ) .
flux calibrations were performed with hz 44 , but the night was not photometric .
figure 1 shows the observed spectrum . within our sample ,
3c 382 is exceptional in several respects .
f@xmath7 rises strongly to the blue , the narrow [ o iii ] lines are very weak , and the broad lines are almost as wide ( at least 26,000 km / sec fwzi ) as the widest found in quasars ( @xcite ) .
we can compare our flux values with those of @xcite who observed the h@xmath0 region of 3c 382 with a @xmath112 slit on 6/7 june 1995 .
their value for the continuum flux at 6500 ( observed ) is about 7 mjy ( figure 1 in corbett et al . ) .
our value is 30 mjy .
corbett s value for the total flux at the peak of h@xmath0 ( actually , the valley between the two peaks ) is about 23 mjy ; ours is 92 mjy .
this comparison is difficult because the shape of h@xmath0 is known to be variable , and also the result will depend on seeing , atmospheric transparency , and on the relative amounts of nuclear and extended light that are in the extracted spectrum . in spite of these concerns ,
the comparison does suggest that in may 1996 both the continuum and h@xmath0 were stronger by a factor of about four than they were in june 1995 .
we can also make a comparison at h@xmath32 by using the results of @xcite .
osterbrock et al .
measure 114 for the total equivalent width , ew , of h@xmath32 and @xmath113 erg @xmath114 sec@xmath115 for the total flux in broad h@xmath32 .
we measure ew=89 and f@xmath116 erg @xmath114 sec@xmath115 .
the continuum and broad lines appear to have been exceptionally high in may 1996 , about 4 times higher than their values a year earlier and about 10 times higher than their values in 1974 .
our observed flux at 5530 is 26.3 mjy , which gives @xmath117 .
3c 382 may even have been the brightest quasar in the sky at that epoch . in ned 3c
382 has flux values ranging from 1.6 to 7.3 mjy at 5530 , so we saw 3c 382 in april and may 1996 at a historic high . correcting to rest wavelength and using 0.3 mag for the galactic extinction gives @xmath118 . in its brighter phases
3c 382 is a highly - luminous quasar . in the b - band polarization image ( figure 4 )
the outer vectors are somewhat circumferential and probably are artifacts due to the strong flux gradient and variable seeing .
the polarization values of the central pixels are unreliable due to saturation . in figure 2a
@xmath15 rises to the blue , as is common in quasars ( @xcite ) .
this is intrinsic and not due to decreasing dilution by galactic starlight , as the starlight fraction is @xmath119 ( table 3 ) .
the broad lines are polarized substantially less than the continuum , consistent with the hypothesis that we see the continuum source and the blr more - or - less directly ; the former is polarized by local scattering , with possibly a synchrotron contribution , and the latter is weakly polarized by scattering in the blr itself or farther out . in our work
the blrg @xmath2 s generally show little correlation with the radio axis ( see 10 ) , but 3c 382 is the one closest to parallel , with @xmath120 .
this might represent a case of scattering in a disk ( antonucci 1984 ) .
the polarization of broad h@xmath0 has been studied by @xcite .
although the flux of both the continuum and h@xmath0 rose substantially between their measurements and ours , the polarizations are similar .
we measure @xmath121 in the continuum at 6500 and 7500 , respectively , and @xmath122 at the bottom of the dip in @xmath15 caused by dilution with h@xmath0 .
corbett et al .
( their figure 1 ) show @xmath123 .
we have @xmath124 with no change at h@xmath0 , while corbett et al .
have @xmath125 with perhaps a small change in h@xmath0 .
( corbett et al .
decompose the continuum and h@xmath0 but here we give values appropriate to the total flux . )
the @xmath2 s are the same but our values for the continuum @xmath15 are lower by a small amount .
this difference may be significant , but the major result is that the polarization changed little while the flux , both continuum and h@xmath0 , increased substantially .
this suggests that the geometry did not change and shows that there is no appreciable starlight even at the fainter levels , for otherwise the polarization would have increased with the decreasing dilution .
figure 7 shows that the balmer decrement and spectral index are consistent with little or no dust along the los to the continuum source and the blr .
the polarized flux ( figure 1b ) has @xmath126 , and is substantially steeper than the total flux .
@xmath127 is not an achromatic version of f@xmath7 but has been bluened .
the image of fsc 2217 + 259 in figure 4 shows a bright star @xmath128 e , and there is another @xmath129 se .
these stars have polarizations @xmath130 at @xmath131 and @xmath132 at @xmath133 , respectively .
this might well be due to isp , and so the observations could be corrected for the average , @xmath134 at @xmath135 .
the correction is small , @xmath136 in @xmath2 , and does not materially change either the image or the spectra .
we do not correct figures 1 and 2 for the isp . recall that in 4.3 we noted that @xmath137 is flat for fsc 2217 + 259 , and that this implies that a substantial foreground polarizing agent is unlikely to exist .
the polarization is high in fsc 2217 + 259 , rising to nearly 9% at 4000 in @xmath15 , and to 13% in @xmath31 .
figure 2a shows that h@xmath0 and h@xmath32 are polarized more than the neighboring continuum , and @xmath15 dips in narrow h@xmath0 . in this respect fsc 2217
+ 259 is like 3c 234 and other nlrg , but the effect is much weaker and for later discussion we will assume that in fsc 2217 + 259 the broad balmer lines are polarized the same as the continuum .
the galaxy - subtracted polarization ( figure 3b ) is like that of 3c 227 ; @xmath138 is reduced in the narrow [ o iii ] lines and is nearly the same in h@xmath0 and the continuum .
fsc 2217 + 259 , however , does not have the @xmath2 rotations in h@xmath0 that are displayed by 3c 227 .
it is unlikely that the polarization in fsc 2217 + 259 is due to extinction by aligned dust grains , for two reasons .
first , @xmath15 has no sign of the serkowski spectrum , which is gently curved with a peak , usually in v - band ( @xcite ) .
secondly , the high value of @xmath15 , 9% at 4000 , is extreme although not unprecedented in stars close to the galactic plane .
vi cyg 12 , at @xmath139 , has polarization similar to that of fsc 2217 + 259 , but is much redder .
it is more likely that the polarization in fsc 2217 + 259 is due to scattering .
figure 7a shows that the high balmer decrement can be explained with 3.8 magnitudes of extinction , but this one - component model leaves us with a very blue intrinsic spectral index , and says nothing about the high polarization . the two - component model in figure 7b , with @xmath140 and @xmath141 yields @xmath142 and @xmath143 . with these parameters
the calculated apparent polarization rises from 4.1% at 7000 to 16.4% at 4000 . as with 3c 227 ,
the calculated polarization is steeper than the measured curve in figure 3b . to match the observations the product @xmath106 would have to be raised by a factor of about 5
this seems unacceptably high .
the model gives a qualitative sense of the spectrum and polarization but does not match fsc 2217 + 259 in detail .
the @xmath144 magnitudes of extinction on the direct ray means that if this object were seen along a dust - free direction , it would have an absolute magnitude m@xmath145 .
further , the broad - line ratio h@xmath146 drops from 10.2 to 4.3 , closer to case b , upon dereddening .
all this is consistent with a quasar seen through dust .
we do not have a polarization image of 3c 445 .
the spectropolarimetry was taken at palomar and the noise level is higher than for the keck data .
3c 445 is like 3c 382 in having very little starlight ; but is different in having strong narrow lines and a redder continuum .
the polarization rotation in broad h@xmath0 is discussed in 9 .
the polarization of the h@xmath0 region has also been recently discussed by @xcite .
our values for f@xmath147 ( july 1994 ) agree well with theirs .
3c 445 has @xmath148 ( ned ) which , with a galactic extinction correction of 0.2 mag , gives @xmath149 .
the internal extinction at v - band , from figure 7 , appears to be between about 1.1 and 2.5 mag . assuming a correction to b - band of 0.3 mag
, we derive an intrinsic @xmath150 between -21.8 and -23.2 mag . from this analysis , based on the balmer decrement and the continuum polarization
, we find that 3c 445 could intrinsically be bright enough to fit into the quasar category @xmath151 .
we first discuss 3c 195 , the prototypical nlrg showing broad balmer lines and spatial symmetry in polarized light . in figure 2b
@xmath2 rotates about @xmath152 between 3500 and 7500 .
this might be due to interstellar polarization , as the galactic latitude of 3c 195 is low and the estimated upper limit on isp is 1.1% ( table 1 ) .
we did not measure stars in the field of 3c 195 , but two stars within @xmath153 have @xmath154 at @xmath155 , and @xmath156 at @xmath157 .
( @xcite ) .
correction for the average , @xmath158 at @xmath159 , yields a curve of @xmath2 that tilts a little in the other direction , indicating perhaps an over - correction .
in addition the [ o iii ] lines become less prominent in @xmath2 than they are in figure 2 , and a @xmath2 rotation in broad h@xmath0 becomes noticeable .
the partial disappearance of the [ o iii ] lines is a sign that the correction may be useful , but the appearance of broad h@xmath0 in @xmath2 seems like an artifact .
altogether , we believe that a correction is needed , one rather smaller than the average shown above . however , given the notorious gradients of isp in the galaxy , we do not pursue this farther , since it needs the measurement of stars in proximity to 3c 195 .
the corrected value of @xmath2 is probably close to @xmath93 .
the correction would have negligible effect in b band because @xmath15 is high , and figure 5 is unaffected by it , except in the regions where @xmath15 is low
. the magnitude @xmath15 ( figure 2 ) has negligible correction , at all wavelengths except the [ o iii ] lines , because the correction vector ( in @xmath160 space ) is nearly perpendicular to the measured vector .
figure 5c shows the b - band polarization image .
the image shows oppositely directed fans " of polarization with vectors that are roughly azimuthally directed .
this is the sign of a reflection nebula , in which a point source illuminates the surrounding material and the scattered light is polarized perpendicular to the plane of scattering .
the fans are roughly co - spatial in projection with the eelr ( @xcite ) .
the general interpretation of this pattern is that an anisotropic beam of radiation illuminates and photoionizes the surrounding material in two oppositely directed cones " .
the eelr fans are the projected cones of ionization .
the beam is scattered by electrons and dust in the eelr and we see a partially polarized version of the nuclear spectrum
. the narrow lines are recombination radiation that is less polarized than the nuclear light because it arises outside ( or perhaps partly inside ) the scattering region .
note that the cones are not symmetric . to the nw
there appears to be a non - illuminated region that extends in close to the nucleus .
to the se the corresponding region has polarized light , close to the nucleus .
this may reflect different s / n on the two sides , rather than a real lack of scattered light nw . in figure 5c @xmath15 is weak in the center ( 3% ) , rises rapidly for @xmath161 to the ne ( 15% ) and sw ( 11% ) , and then decreases slowly .
the strong gradient is somewhat surprising , since a simple model , consisting of optically thin scattering from electrons or dust with density decreasing linearly with radius , gives @xmath15 that is independent of radius .
figure 5d shows a contour diagram of the polarized flux , @xmath162 , whose axis is oriented at pa@xmath163 , rather far from the radio axis at @xmath164 .
@xmath17 has two peaks ; the stronger one is displaced about @xmath165 sw of the flux peak , and the other is @xmath161 ne , where the flux contours are extended and bend n. the reduction in @xmath15 and @xmath17 at the center is discussed in 5.2 , where we suggest that it is due to the cone geometry , with dilution by starlight also affecting @xmath15 .
this reduction in @xmath15 and @xmath17 is common , as it exists in 4 of the 5 highly polarized nlrg .
a uniformly filled scattering cone will show a polarization at the apex that is perpendicular to the axis of the system . in the central @xmath166 of figure 5c
the vectors have @xmath167 , and we assume they are perpendicular to the axis of the cone , which then is at pa @xmath168 . the axis of @xmath17 ( 35 ) is close to this cone axis . a line at @xmath169 is within about 2of perpendicular to all the vectors along its path .
this line is the symmetry axis of the reflection nebula , because of the perpendicular polarization at the apex , and also the close perpendicularity along the line which is not matched elsewhere . the axis , and the rotation in @xmath2 , can be appreciated in figure 8 , which is a color - coded diagram of @xmath2 .
the rotation is clear in the north , where the colors are roughly radial , and a center can approximately be located where the various sectors come together .
this point is the illumination center .
it appears to lie between the peaks in @xmath170 , and is not at the flux peak , but rather is displaced to the ne by @xmath171 .
this situation may be analogous to the displacements in cygnus a , where the center is hidden by dust ( o97 )
. it would be interesting to compare the radio and optical positions for 3c 195 to see if there is an offset , but the optical positional uncertainty is comparable with our measured displacement ( @xcite ) . the pattern in figure 8 is more regular in the n than the s , but the axis can be traced across the entire object . in the e and w
the red regions are roughly perpendicular to the radius , but they do not extend into the central region .
it is likely that the nuclear region is complex , as it is for the galaxy and for cygnus a , with broken dust clouds and , perhaps , multiple scattering .
it is important also to remember that the polarization images are smoothed by the seeing weighted by the spatial distribution of @xmath33 , and so in weakly polarized regions @xmath2 will be distorted towards its value in nearby strong regions .
note that the value of @xmath2 obtained with slit spectroscopy may depend on the location of the slit and on the extraction window .
in general , the vectors are close to perpendicular to the radius , and so , away from the nucleus , the measured @xmath2 will be close to perpendicular to the slit orientation .
this means that one must take care in comparing a polarization to a radio axis in an nlrg like 3c 195 , because the value of @xmath2 will depend on the details of how it is measured . for spectropolarimetry we oriented the slit to include the strongly polarized regions shown in figure 5d .
we extracted the spectrum and polarization in the nucleus and in regions to the ne and sw , but in this paper we only present spectra centered on the nucleus .
subtraction of the galactic light is not enough to flatten @xmath138 ( figure 3 ) .
the corrected polarization rises to the blue and is about @xmath172 at 3500 .
also , broad h@xmath0 and h@xmath32 are polarized more strongly than the continuum ; this might be due to a second continuum component , fc2 , as discussed in 4.2 ; see also 8.2 .
3c 33 , a well - studied fr ii radio source , is associated with a galaxy having an eelr that extends far outside the galaxy .
the velocity field in the eelr has been mapped by @xcite , and an optical synchrotron emission region is coincident with the southern hot spot ( @xcite ) .
the main h@xmath0 axis is not close to the radio axis ; this is a case where the alignment effect is weak .
our b - band image is in figure 5a .
a few emission lines , notably [ o ii]@xmath51 and [ ne iii]@xmath173 , are in the b - band flux , but the lines amount to no more than 25% of the total and the image is a good representation of the continuum .
the lines themselves are essentially unpolarized and do not contribute to the polarized flux .
the b - band contours are elongated in the central few arcseconds , with an axis parallel to that of the h@xmath0 image .
they become circular outside a diameter of @xmath174 , so it is only close to the nucleus that the continuum displays the elliptical shape , whereas the h@xmath0 image is elliptical over its total extent , about @xmath175 . the polarized regions in figure 5a are asymmetrical , but close to the nucleus they lie roughly along the continuum and h@xmath0 axis .
the ne polarization region is much larger than the one to the w. the polarization shows a strong gradient , with a peak value 8.4% in both the ne and the w. the polarization is low near the nucleus , where @xmath176 within the central arcsecond .
this value is low enough that it could be affected by isp , but the high galactic latitude ( -49.3@xmath177 ) suggests that the effect is small .
three stars separated from 3c 33 by 3.5 , 5.5and 6have polarizations 0.2% , 0.4% , and @xmath178 , respectively ( @xcite ) .
it is likely that the isp in the direction of 3c 33 is a few tenths of one percent , but we ignore it because ( a ) in fact we do not discuss the polarization at the nucleus in detail ( see 5.2 ) , and ( b ) we can not correct for it without observations of stars close to 3c 33
. the weak isp does not affect the high polarization outside the nuclear region .
the plot of @xmath179 in figure 5b is closely similar to that of 3c 195 in figure 5d .
both objects show separated peaks differing in intensity by approximately 30% , and a low - level extension on one side . the position angle
@xmath2 rotates around the nucleus in figure 5a and stays roughly perpendicular to the radius .
thus 3c 33 provides another good example of a reflection nebula illuminated by a point source . in the nucleus @xmath180
this is close to parallel to the @xmath179 axis rather than being perpendicular to it , as in 3c 195 .
3c 33 does not contain a clear polarization axis , as 3c 195 does .
the radio axis , however , is at @xmath164 which , like the axis in most of the nlrg , is within @xmath95 of perpendicular to the nuclear @xmath2 ( see 10 ) .
draper , scarrott , & tadhunter ( 1993 ) report imaging polarimetric observations of 3c 33 in v - band , which contains strong [ o iii ] lines and h@xmath32 .
the narrow lines are polarized less than the continuum , however , so the draper et al .
polarization image is dominated by the continuum and can be compared with our b - band image .
the results generally are similar , and we confirm the centro - symmetric pattern they report . however , we find a stronger gradient , with low polarization at the nucleus .
spectropolarimetry of 3c 33 was made with the slit along the major axis of the h@xmath0 image of baum et al .
( 1988 ; pa @xmath181 ) as shown in figure 5b . in figures 1 - 3 we show spectra extracted from the region of the ne peak in @xmath179 , 2@xmath182 from the nucleus .
this region has @xmath183 , nearly perpendicular to the slit direction , as expected for a resolved reflection nebula .
3c 33 is an nlrg , but a weak broad line can be seen under h@xmath0 in figure 1a , and broad h@xmath0 is the dominant feature in polarized light , figure 1b . the starlight fraction is high ( table 3 ) and the corrected flux in figure 3 shows prominent broad balmer lines .
the polarization corrected for starlight is shown in figure 3b . in broad h@xmath0
@xmath138 rises to @xmath184 ; this is the highest value of @xmath138 that we have measured .
this perhaps is not surprising , as it is for an off - nuclear location where geometric dilution is expected to be smaller than near the nucleus .
the continuum polarization is lower , near 12.5% , suggesting that there is a strong fc2 component .
the position angle shows no changes at h@xmath0 so we can assume that the nuclear continuum and h@xmath0 are intrinsically polarized the same . in this case we can follow the analysis of tran ( 1995 ; see eqs 2.5 - 2.7 ) , and we calculate @xmath185 , and the intrinsic polarization of fc1 and broad h@xmath0 , @xmath86 , is about 50% . that value is basically calculated at one wavelength , h@xmath0 , where @xmath138 is a maximum . to obtain a feel for the validity of our analysis , we plot in figure 9
the spectrum of fc2 ( the unpolarized flux spectrum " , see tran 1995 , figure 1 ) with @xmath86 held constant in wavelength .
the top spectrum shows the corrected flux ; this is the same as the plot in fig .
the lower curves show spectra for @xmath186 . the fit can best be examined in the blue wing of h@xmath0 , at 6500 .
it is clear that @xmath187 undercorrects for fc2 , and @xmath188 overcorrects .
a value between 0.4 and 0.6 appears to provide a good average correction over the h@xmath0 line .
the noise is higher in h@xmath32 , but @xmath189 appears to be valid there also .
our value for the intrinsic polarization due to scattering , @xmath190 , is higher than values reported earlier .
this is at least partly due to the off - nuclear position of the measurement , but 3c 234 appears to have no geometric dilution ( fig 5e ) , and for it we found @xmath191 ( t95 ) . in a group of 10 seyfert 2 galaxies tran ( 1995 )
found a maximum value @xmath192 .
high values of @xmath86 , like the one we find in 3c 33 , in fact are expected in simple scenarios of scattering and the low values more commonly found have required some explanation ( e.g. @xcite ) .
3c 105 is the only object in our sample that shows no evidence for polarization , with @xmath193 .
the galaxy is exceptionally red , with strong stellar absorption lines ( @xcite ) .
the keck spectra are shown in figure 10 ; the bottom panel is the spectrum with a galactic template ( corrected for reddening ) subtracted .
there is no reliable indication of broad lines .
3c 105 has a typical fr ii radio morphology , with a weak core and prominent hot spots , and is similar to 3c 357 .
3c 135 has several nearby companions ( figure 6 ) , including one @xmath194 sw and a highly - distorted one @xmath195 sw .
the hst image ( @xcite ) also shows a fine - scale extension from the nucleus to the sw .
the polarization of 3c 135 is low ; even after the strong correction for host galaxian starlight it is only about 3% , with a large error .
eracleous & halpern ( 1994 ) did not find a broad component of h@xmath0 in 3c 135 , and h96 similarly found narrow but not broad p@xmath0 .
however , although 3c 135 is weakly polarized , broad h@xmath0 and ( perhaps ) h@xmath32 are visible in @xmath15 and @xmath196 .
3c 321 ( table 4 ) is similar in not showing broad p@xmath0 while showing broad h@xmath0 in polarized light . in the polarization image of 3c 135 ( figure 6 )
the vectors are stronger than @xmath197 in 10 bins near the nucleus .
the value measured near the center is @xmath198 with @xmath199 .
the image is taken in b band that contains [ o ii]@xmath51 plus starlight in addition to the weak continuum .
the spectropolarimetry is noisy in b band but at 5500 the value is @xmath200 with @xmath201 ( table 3 ) .
the values are consistent with a polarization that rises to the blue , as is common in nlrg , and they are marginally consistent with @xmath2 being independent of wavelength . the spectropolarimetry for 3c 234 has been discussed by t95 .
figure 5 shows contour diagrams of f@xmath7 and @xmath18 .
@xmath15 increases to the east and has a maximum about @xmath202 east of the flux maximum .
this gradient is compensated by the fall of f@xmath7 , so that @xmath18 and f@xmath7 have maxima in the same pixel .
figure 5f shows that the east - west asymmetries in f@xmath7 and @xmath15 combine to produce @xmath203f@xmath7 that is close to symmetric , although it is noticeably thinner north - south than is f@xmath7 .
the spatial distributions of @xmath15 and @xmath18 in 3c 234 are different from those in both the blrg and the other nlrg .
the polarization pattern is extended but does not show fans , and at the nucleus @xmath15 is neither a minimum nor a maximum .
the pattern is probably due to scattering of the nuclear continuum , but the geometry seems to be different from that of the other objects .
the lack of polarization fans suggests that we are looking inside the cone , but we must be close to the edge because the broad lines are enhanced in polarized light ( much more so for 3c 234 than fsc2217 + 259 ) .
it is hard to explain the symmetry in @xmath203f@xmath7 in this case , especially the lack of circumferential vectors to the north and south .
the source @xmath204 ne ( figure 5e ) was recognized by cimatti & di serego alighieri ( 1995 ) who stated that it is highly polarized at v band . in our b - band polarization image
no bin in that region shows @xmath205 .
we integrated over a @xmath206 square on the ne peak and found @xmath207 .
hurt et al .
( 1999 ) , using hst ultraviolet imaging polarimetry , also fail to detect polarized light from this region , and suggest that it is dominated by an old stellar population , based on the red color .
the b - band polarization image of 3c 321 is shown in figure 5 g .
the gradient in @xmath15 is stronger than it is for the other objects ; and , indeed , in the central region no individual bin has a measurable polarization stronger than @xmath208 .
the polarization is marginally detected in a 4x4 bin @xmath209 ) average in the center : @xmath210 .
@xmath15 rises to @xmath211 in the nw and to @xmath212 in the se .
the vectors are roughly circumferential , indicating scattering of light from a point source .
this pattern has already been noted by draper , scarrott , & tadhunter ( 1993 ) . in an attempt to find the source of the nuclear light
, we analyzed the angle @xmath213 between the local value of @xmath2 and the radius to an assumed center .
this procedure has been used before by @xcite . when applied to 3c 321
, it shows that @xmath214 is a minimum at the se flux peak , where it has the value @xmath215 , whereas at the nw peak it is @xmath216 . in other words ,
the vectors point ( perpendicularly ) more to the se component than to the nw component , and the source of nuclear light appears to be in the southern component .
this is consistent with filippenko s ( 1987 ) result that the compact radio source is in the se component .
3c 321 appears to be different from cygnus a , where the center of activity is hidden by dust and lies between the main nuclear components .
close inspection of the polarization vectors shows that deviations are not random , but rather that the vectors are systematically tilted away from the se nucleus .
this can be caused by seeing , combined with the spatial gradients .
young et al .
( 1996 ) found broad h@xmath0 in the polarized spectrum of 3c 321 and , by using the balmer decrement to estimate the reddening , showed that the intrinsic luminosity was sufficient for the nucleus to harbor a hidden quasar .
the spectra we show in figs 1 - 3 were made with the slit along the optical symmetry axis and with an extraction window @xmath217 long , that includes essentially all the galaxy seen in figure 5 .
broad h@xmath0 is weakly visible in the total flux in figure 1 , but is prominent in the polarized flux and in the galaxy - corrected flux ( figure 3 ) .
broad h@xmath32 and h@xmath218 also are visible . figure 5 shows that the polarization varies widely over the galaxy , so the measured values of @xmath15 and @xmath2 will generally depend on the slit location and the extraction window .
the rapid rise of @xmath15 to the blue in figure 2 is partly but not completely due to dilution by an old stellar population , as the increase of @xmath31 in h@xmath0 shows that fc2 is present also .
@xcite attribute this extra continuum component to a stars .
the spectrum of polarized flux is steep in 3c 321 , with @xmath219 ( figure 1a ) .
this is the steepest of the nlrg , and is as steep as 3c 382 , which also has @xmath219 . in the uv
the polarization rises to 40% ( @xcite ) .
h96 did not detect broad p@xmath0 in 3c 321 , but in our observations the broad lines are obvious in @xmath220 ( fig .
1 ) . this shows the importance of complementary observations .
we discuss this further in 10.2 .
the hst image of 3c 357 ( @xcite ) shows dust lanes and a patch of emission about @xmath161 w of the galaxy ; we call this region w. on our keck b - band image ( figure 6 ) w is seen as the bulge in the second - brightest contour ( 20% ) .
it lies close to the radio axis .
the nucleus does not show any polarization , but in w four vectors survive the @xmath197 cutoff , giving @xmath221 .
the vectors are closely perpendicular to the radius , suggesting a scattering origin for the polarized light .
the spectroscopic slit at @xmath222 was centered on the polarized region , which is near the southern edge of region w. both the nucleus and w show a great deal of starlight ( table 3 ) and in figure 3 we show the galaxy - subtracted flux and polarization of w. ( the flux is shown on a linear scale because it is so noisy . )
the narrow lines of w are red - shifted with respect to the nucleus , by 607 km / sec .
imaging polarimetry of w gives 4% in b band and spectropolarimetry gives @xmath223 at 5500 .
the rise to the blue is due to dilution by starlight , although narrow lines in the b band image should reduce the b band polarization .
the galaxy - corrected polarization @xmath224 in the continuum .
the noise is low in the strong narrow lines [ o iii]@xmath50 and ( h@xmath0 + [ n ii ] ) and @xmath225 drops to near zero in the lines , indicating that they are polarized much less than the continuum , and arise _ in situ_. from the nuclear spectra ( not shown ) , an upper limit for @xmath226 in the nucleus is 0.4% ( @xmath197 ) .
3c 357 is an unpolarized nlrg like 3c 105 ( 8.3 ) , but contains a fortuitously located scattering region in the form of region w. it may be similar to the off - nuclear scattering region in ngc 1068 ( @xcite ) .
there is no evidence of a blr in 3c 357 , although our s / n is poor
. it would be worthwhile to repeat these observations , to establish more clearly the presence or absence of broad balmer lines .
this is of importance in understanding the unpolarized nlrg ; is the observed lack of broad lines intrinsic , or merely due to a lack of scattering material ? the s / n of the current data is too low for a definitive statement , but the apparent lack of a scattered blr is an intriguing suggestion that there may not be a blr in 3c 357 .
cygnus a , the prototypical nlrg , is discussed by o97 . the fact that it has a reflection nebula and shows broad lines in scattered light gives strong support to the unification by aspect of blrg and nlrg .
figure 2b shows intriguing rotations in @xmath2 across broad h@xmath0 in 3c 227 and 3c 445 ; there are smaller rotations in 3c 321 and cygnus a , and similar effects in other lines in 3c 227 .
similar rotations have been seen in seyfert 1 galaxies ( @xcite ) and in broad - absorption - line qso ( @xcite ) . evidently , such rotations are common in broad - line objects . in this section
we describe the rotations in 3c 227 and 3c 445 .
figure 11 gives an expanded view of the polarization of 3c 227 , showing ( a , b ) the stokes parameters @xmath227 and @xmath228 , and ( c ) the position angle @xmath2 . to study the pa rotations in more detail we decompose the polarized radiation into continuum and line components , with @xmath229 and @xmath230 .
@xmath231 and @xmath232 are approximated by the smooth curves in figures 11a and 11b ; they are second - order polynomial fits to @xmath233 and @xmath234 in regions free of emission lines .
the resulting @xmath235 is the smooth curve in figure 11c , which is shown together with @xmath236 . in figure 11d
the smooth curve is again @xmath235 and the other is @xmath237 .
we emphasize that this construction depends critically on the exact curves used for @xmath231 and @xmath232 , and a reliable decomposition can only be made with high s / n data . in the spectrum of 3c 227 ( fig 1 ) a weak broad feature is visible at 5700 - 5900 .
this is a blend of he i@xmath238 , na i@xmath239 , fe vii@xmath240 and , probably , other lines .
( see thompson 1991 for a discussion of the he and na lines . )
this feature is polarized and is seen in the curves of @xmath241 and @xmath2 in fig .
similarly , the blue wing of h@xmath218 contains a contribution from broad he ii@xmath242
. we shall refer to these blends simply as he i@xmath238 and h@xmath218 . in figures 11c and 11d
it is clear that broad h@xmath0 , h@xmath32 , he i@xmath238 , and possibly h@xmath218 , have a pa that is different from that of the continuum .
h@xmath0 and h@xmath32 are rotated about @xmath95 from the continuum , and @xmath2 is constant in these lines to within about @xmath243 . h@xmath218 and he i are rotated from the continuum in the same direction as h@xmath0 and h@xmath32 and possibly by the same amount , although they are weak and the errors , due mainly to uncertainties in the continuum levels , are large .
the narrow lines are polarized less than the continuum and broad lines , as is usual in these rg .
[ o iii]@xmath244 appears as a prominent spike in @xmath2 , and formally is at @xmath245 , but we are suspicious of this value , because it occurs in only one pixel .
it might be due to noise , or to problems with the spectrum extractions .
[ s ii ] appears to show up in @xmath2 and @xmath246 , but this may be due to noise as the [ s ii ] signal is weak in @xmath241 and @xmath247 .
figure 12 shows the h@xmath0 region of 3c 445 in the same construction as for 3c 227 , except that we used a linear fit to the continuum for @xmath233 .
note that in ( c ) and ( d ) the scales differ by a factor of two .
the rotation in h@xmath0 shows a striking red / blue difference in 3c 445 , one lacking in 3c 227 .
the crossover occurs at about 6500 , which is offset from the narrow - line systemic velocity by 3000 km / sec .
the red side in @xmath248 ( figure 12d ) is roughly constant at @xmath249 and is rotated by about @xmath243 from @xmath235 .
this component clearly persists to 6820 or to the end of the red wing in total flux .
its total velocity width is 15,000 km / sec .
the blue side may persist to 6180 , although the noise near 6300 makes this questionable .
its width is at least 7000 km / sec and could be up to 15,000 km / sec .
the value of @xmath248 clearly changes across the blue component , from about @xmath250 to about @xmath251 .
the peak difference between the red and blue components is about @xmath252 .
the two values of @xmath248 for h@xmath0 suggest that we are dealing with two components of the radiation , that are separated in velocity and have different pa s .
the blue component ( b ) is dominant below 6500 , and the red component ( r ) is dominant above 6500 .
inspection of the @xmath179 curve ( figure 1b ) shows that h@xmath0 has a pronounced break near 6500 , and so such a decomposition seems natural .
b and r are separated in velocity by @xmath253 km / sec and are polarized @xmath254 with @xmath2 s that are @xmath255 apart .
the seyfert 1 galaxies studied by goodrich & miller ( 1994 ) , martel ( 1996 ) and smith et al .
( 1997 ) show a variety of rotations in the broad lines , and several of their objects show rotations that are as large and as complicated as those in 3c 445 .
these authors discuss this phenomenon in terms of two or more components that we see via different scattering regions .
the regions must lie fairly close to the blr , for the pa s to differ substantially .
martel decomposes f@xmath7 into gaussians , and the best case is ngc 4151 which yields 7 broad components .
he then identifies some of the components with structures in the galaxy ( a bar , radio and optical axes ) on the basis of pa coincidence .
we do not attempt a similar deconstruction for 3c 227 and 3c 445 here , but restrict ourselves to one component of h@xmath0 for 3c 227 , and two for 3c 445 . in 3c 227 and 3c 445
none of the pa s is close to being either parallel or perpendicular to the radio axis .
the polarization does not result from simple scattering on a distribution that is symmetric around the radio axis .
what then is the source of the polarization ? in 3c 227 we evidently have different lines of sight into the nuclear continuum source and the blr .
these light sources are spatially separated by a few tenths of a parsec , so we can imagine that different scatterers are responsible for the two different polarizations .
the los is not close to the axis of the dusty torus because the radio source has fr ii morphology with a weak core , and the inclination is probably many tens of degrees .
thus , the required asymmetry is plausible .
a fraction of the h@xmath0 light could be scattered and polarized by dust clouds in or near the blr , and it is possible that absorption in aligned dust particles in front of ( or mixed with ) the h@xmath0 clouds is responsible for some of the polarization .
the dust alignment need not have any simple relation to the polarization of the continuum , which is set much deeper inside the nucleus .
3c 445 is more complex , in that we must account for two components of h@xmath0 , but that also offers more constraints and may limit the range of possible models .
there is not enough information to invert the polarimetry data and solve for the location and motion of the h@xmath0 clouds .
however , we can investigate some simple limiting cases and try to find plausible situations .
we assume that the system contains a blr inside a dusty opaque torus , and that the continuum source is at the center .
the los skims the edge of the torus and we have a limited view of the interior .
we first note that the motions of the h@xmath0 clouds can not be chaotic , since then the red and blue shifted sources would be spatially mixed , and the red and blue components would have the same pa , contrary to observation .
hence the clouds must be in streaming motion .
this assumes that the velocity shifts are due to motions of the emitting clouds , and not of the scatterers .
further , the lines can not be broadened purely by thermal motions in clouds , for in that case the electrons would produce equal red and blue shifts at every velocity .
blr clouds orbiting inside the torus , near the equatorial plane , provide a scenario which can explain much of the observed asymmetry in the polarization of h@xmath0 .
consider the case of clouds orbiting in circles in the equatorial plane .
the motions are not aimed at us , so the observed high velocities must be due to scattering by clouds that see the full red and blueshifts . in other words ,
the polarized h@xmath0 light that we see has been scattered by material that is close to the orbital plane .
this material can form the inner wall of the torus . if the rotating blr clouds are optically thick they mainly radiate back towards the ionizing source , but still , for circular orbits , half of each emitting hemisphere is visible at the rim and the full range of velocities can be seen in the scattered light .
an appropriate combination of inclination angle and shape of the inner wall of the torus can now provide red and blue components that have different pa s .
see an example in figure 13 . in this example
all clouds move on circles in the same direction . in discussing mrk 231 ,
goodrich & miller ( 1994 ) describe a scenario that is similar to the one we present here .
our scenario only works over a restricted range of inclination angles .
if the inclination is low the nuclear region is entirely visible and we see a quasar . at high inclination
we see an nlrg with polarization perpendicular to the axis .
if the view is near the boundary of the torus the equatorial plane is partially visible ( fig 13 ) and we see a blrg with different pa s for the red and blue wings of h@xmath0 .
this simple model can not be taken literally , for the situation undoubtedly is more complex .
for example , @xcite have described a scenario , based on x - ray spectroscopy , in which the blr is in a warped disk that is not coaxial with the accretion disk .
however , we believe that our concepts still have some validity .
the clouds are in streaming , not chaotic motion .
optically thick clouds in equatorial orbits provide a geometry that meets the observations , although the 3000 km sec@xmath115 offset to the blue is not readily accounted for .
reverberation studies in seyfert 1 galaxies ( e.g. @xcite ) show that circular orbits are preferred over radial orbits , and this provides an extra measure of support for our scenario .
the model in 6 describes the blrg radiation as a combination of two components , a reddened direct ray and a polarized and reddened scattered ray .
the relative amounts of the two rays fix the spectral index and the amount and spectrum of the polarization .
the nlrg have more attenuation of the direct ray , so that the nuclear continuum and broad h@xmath0 are not directly seen .
they show up in the scattered light , and a weaker component , fc2 , also becomes important .
the evidence for this is that @xmath31 rises in h@xmath0 , indicating a diluting unpolarized component in addition to the polarized continuum fc1 .
observationally , the blrg can be divided into 3 groups that have rather different spectra and polarization .
however , these should not be thought of as separate classes of objects , since they are primarily distinguished by differing extinction and inclination angle .
@xmath256 3c 382 is very blue , has low polarization , and broad h@xmath0 is prominent in total flux but not in polarized flux . in its brighter phases ,
this object is a quasar .
the low polarization of h@xmath0 can be understood if we have a direct , unattenuated view of the blr , so that the scattered ( polarized ) light from the blr is weak in comparison to the direct ray .
another way of looking at this is to imagine that all blrg have a more - or - less standard spectrum of @xmath220 , as exemplified by 3c 227 .
3c 382 additionally has a strong unpolarized component of h@xmath0 that dilutes the polarization in the line .
this combination leaves the position angle unchanged in the line , as required by figure 2b .
@xmath257 3c 227 and 3c 445 are highly polarized in the blue .
the broad lines are equally prominent in total and polarized flux .
pa rotations in h@xmath0 show that there are multiple lines of sight to the blr .
we have a view deep into the illuminated cone , but may not directly see the nuclear continuum source or the blr .
the view probably is close to the edge of the torus .
these two objects are analogous to the type 1 seyferts that have pa rotations , but the rg also have powerful radio jets .
@xmath258 3c 109 and fsc 2217 + 259 are very red , highly polarized , and the polarization of broad h@xmath0 is the same as that of the continuum .
they are closely similar except for the spectral index of the continuum ( figure 7 ) .
fsc 2217 + 259 is a quasar seen through about 3 magnitudes of dust extinction ( according to 7.4.1 ) , with a scattered component in which the continuum and broad lines are polarized the same .
the net polarization @xmath259 is high and rises strongly to the blue ; this is due to the reduction in dilution by the heavily - reddened direct component .
3c 109 has been interpreted as being polarized by selective absorption in aligned dust grains , but in view of its similarity to fsc 2217 + 259 we think it may also have a scattered component .
observations into the uv would be useful for both these objects . in any event ,
3c 109 and fsc 2217 + 259 are quasars seen through dust .
we are looking deep into the torus , and may have a direct ( but attenuated ) view of the blr and the nuclear continuum source .
fsc 2217 + 259 does not have a compact optical nucleus like the other blrg , and its image of @xmath18 resembles that of 3c 234 .
our work strongly supports the hidden nucleus hypothesis for nlrg and the unification by aspect of many fr ii nlrg , blrg and quasars , in four ways .
@xmath256 reflection nebulae , in the form of fans of polarized light , are seen in some nlrg .
the polarization vectors are circumferential and accurately define a central hidden continuum source .
the fans are roughly co - spatial with the eelr , and the radio axis is usually inside the fans .
@xmath257 broad h@xmath0 is seen in 6 of the 7 polarized nlrg , showing that they contain a hidden blr .
our high - quality spectra show this feature directly , but it is much more prominent in polarized light . the seventh polarized object , 3c 357 , is the weakest and is unreliable because the atmospheric a band falls on the red wing of h@xmath0 . @xmath258 the intrinsic luminosity of the blrg in our sample , and also of 4 nlrg , 3c 227 ( prieto et al .
1993 ) , 3c 234 ( t95 ) , 3c 321 ( @xcite ) and cyg a ( o97 ) , has been investigated in detail .
it is sufficient to put all of them in or close to the quasar class .
@xmath260 in the blrg the outer reflection nebulae must exist , as they do for the nlrg , but they have not been seen .
however , in blrg the nuclear continuum and broad lines swamp the weaker light from the reflection nebula .
a measure of this difference can be gained from table 3 , where the galaxy fractions @xmath261 show that the starlight ( and presumably the outer scattered light ) is weaker in the blrg than in the nlrg by a factor of 3 to 10 or more .
a further reason why the reflection nebulae might be hard to see in the blrg is foreshortening due to the smaller inclinations .
although the highest - redshift object is 3c 109 , a blrg , the other blrg have redshifts in the range of the nlrg , and angular size is not a major factor in the visibility of the eelr .
another commonly - used aspect indicator is r , the ratio of the radio flux density in the core to the lobe flux density .
log r at 6 cm wavelength is listed in table 5 .
the mean value of log r for the blrg is 1.1 and for the nlrg it is 2.1 .
the difference is highly significant and shows that the blrg radio axes are closer to the los than the nlrg axes , on average .
these results justify the common scenario of an occulting torus that hides the nucleus from direct view but allows collimated nuclear light to be scattered into our los . from other aspects
the nlrg would look like blrg and quasars .
our results suggest that most polarized nlrg contain a hidden blr .
however , we can not generalize about the fraction of unpolarized objects that might show broad lines because we have only one unpolarized nlrg .
we also can not generalize to find the statistical occurrence of broad lines in all nlrg , for the sample is biased towards objects known in advance to be polarized .
broad lines should be seen in the ir if the dusty torus picture is correct .
h96 looked for p@xmath0 in a complete sample of 11 fr ii radio sources , including 1 quasar , 1 blrg , and 9 nlrg .
their sample includes 3c 273 which usually is not called fr ii because it is not a classical radio double .
however , we keep it because it has high luminosity and one visible relativistic jet with an outer lobe
. one nlrg , 3c 319 , is dominated by starlight and has no apparent h@xmath0 ; it was not observed in the ir but we keep it in the sample , which consists of all fr ii 3cr objects in the ranges of flux , redshift , latitude , right ascension and declination chosen by h96 .
we have changed their designation of 3c 234 from blrg to nlrg .
three of their nlrg , 3c 135 , 3c 234 , and 3c 321 , are in our sample .
table 4 summarizes the combined data sets .
y and n simply indicate whether there is or is not evidence for high polarization , broad h@xmath0 and broad p@xmath0 . we have added the polarization value ( n ) for 3c 236 from @xcite , because although a galaxy - subtracted value is not available , the measured value is low , @xmath262 .
table 4 contains 14 nlrg , of which 10 show evidence of broad lines and 4 do not .
there is a strong correlation between polarization and the existence of broad lines : 6 of the 7 nlrg with high @xmath138 show broad lines , and the 2 with no measurable continuum polarization do not show polarized broad lines .
this reinforces the picture seen in our data alone , that polarized objects will show broad lines , either in polarized light or in the ir .
h96 had a complete , if small sample , and the combination of our spectropolarimetric results with their near - ir results is more powerful than either technique alone .
this produces a result that is correspondingly more robust .
while h96 did not see broad p@xmath0 in 3c 135 and 3c 321 , we have shown that there is indeed a blr in these objects .
this brings the number of objects known to have a blr to 8 in the complete h96 sample of 11 fr ii objects , or 6 of the 9 nlrg .
@xcite have made a polarization survey of a complete sample of radio galaxies , of which 6 are fr ii nlrg .
three of the 6 are polarized , but none shows evidence of broad lines .
this seems to contradict our result that polarized nlrg show broad lines , but limits on the strength of the broad lines are not available for their objects .
we believe that a sensitive survey of a larger complete sample is warranted .
the question of whether the unpolarized fr ii nlrg contain a blr is important , and unanswered . in particular , the lack of evidence for broad lines in an object should not be construed as showing that it has no blr .
there may be no signs of broad lines because there is no blr , or because there are no appropriately placed scattering regions .
3c 357 is essentially unpolarized except for a small region ( 3c 357w ) where the hst image shows an off - nuclear emission region .
the polarization there is approximately perpendicular to the line to the center ( figure 6 ) , which suggests that it sees a central continuum source .
it does not appear to scatter any broad lines , but the low s / n does not allow a firm conclusion to be drawn about the existence of a blr in the nuclear region .
3c 105 is unpolarized but still is a powerful fr ii radio galaxy with a strong nlr , and it should have a bright central engine . in this case
the lack of continuum polarization may be due to a lack of scattering material .
the lack of observed broad lines could be due to the absence of a blr , but alternatively it could also be due to a lack of scattering material .
the referee observes that the apparent lack of scattered nuclear light could alternatively be due to obscuration of the scatterers ; this is possible but seems unlikely because the obscuring material would have to cover a large area of the ism .
a polarization image of 3c 105 could be used to search for off - nuclear scattering regions which may have fallen outside our slit .
fosbury points out ( private communication ) that it also is conceivable that the central black hole is not currently being fed , so that the nuclear continuum source and blr are weak , but the long time scales associated with the radio source and the nlr keep them going . as far as we are aware , there are no vlbi observations of 3c 105 , so there is no structural information on scales smaller than a few hundred pc . in powerful rg the radio axis
is assumed to mark the dynamical axis of the central black hole and accretion disk , and it usually is assumed that any obscuring torus has the same axis . in fr ii objects the radio jets are nearly straight and the projected axis is well - defined . in table 5 we give pa for the radio axis , @xmath2 for the optical polarization , and @xmath263 , the acute angle between them .
the value of @xmath2 for 3c 33 is for the nucleus whereas in table 3 it is for the ne lobe , corresponding to the spectrum in figure 1 ; and for 3c 321 it is for the se flux peak ( figure 5 g ) whereas in table 3 it is for the wide spectrum extraction . in the past much attention has been paid to the relation between @xmath2 and the radio pa , and in particular to the relative orientation of the radio axis with the extra - nuclear uv and emission - line light ( alignment effect ) . in 3c 195 , 3c 321 and
cygnus a the radio jet is well inside the polarization fans and the torus axis can be taken as coincident with the radio axis , although the torus is far from a smooth circular structure and the illuminated cones must be irregular . in 3c 33
the polarized fans are not coaxial and the collimation mechanism appears to give different illumination patterns on the two sides , even though the radio structure is nearly straight across the nucleus .
the inner 6 of the @xmath179 image ( figure 5b ) is nearly symmetric but the large weak region to the n is not repeated in the s. in 3c 33 the @xmath179 axis is far ( @xmath264 ) from the radio axis .
the important feature of the polarization image in nlrg is that it shows the reflection nebula . at the nucleus itself
the polarization becomes weak but in all cases it is within @xmath265 of being perpendicular to the radio axis ( table 5 ) .
we think this comes about because , when there is sufficient symmetry at the apex , the contributions add up to give an orientation perpendicular to the cone axis , which is close to the radio axis . in 3c 33
the alignment effect is one - sided , but the nuclear polarization is still roughly perpendicular to the radio axis . in this case
the polarization rotates strongly and is weak at the nucleus ( figure 5a ) so that observations must have high s / n and high angular resolution to be reliable .
@xcite noted that the blrg are mostly polarized with @xmath2 roughly parallel to the radio axis . from table 5
we see that the blrg differences do not cluster near @xmath266 or anywhere else .
however , there are none near @xmath267 , and the median value of @xmath263 for the blrg is @xmath268 , whereas for the nlrg the mean is @xmath269 .
see also hurt et al .
( 1999 ) for nlrg measured in the uv ; they get essentially the same result .
the geometry is simple for the nlrg , because the bright central source is occulted and the scattering takes place far from the nucleus . with the blrg
we have a view deep into the ionization cone , but perhaps not into the nucleus itself .
the polarization is predominantly produced by scattering , and if the ism is irregular a range of @xmath2 is to be expected .
the one object that looks like a quasar , 3c 382 , has @xmath120 , showing that its polarization is close to parallel to the radio axis . in this case
we apparently do have a direct view of the nuclear continuum source , and the various models that have been proposed to explain the polarization , typically involving an equatorial disk , may be appropriate .
the fr ii radio galaxies have a powerful radio jet that is indicative of magnetic field and energetic particles .
hence we might expect that synchrotron radiation could play a role in the optical emission and polarization .
none of our objects shows direct evidence of nuclear synchrotron radiation .
however , we believe that they are obscured or mis - directed quasars . thus if ( radio - loud ) quasars generally do contain a mis - directed blazar , then our objects should also , and there should be a component of synchrotron emission in the continuum radiation . the direct synchrotron beam is narrow and only seen close to the axis .
however , in misdirected objects the beam should illuminate the ism and there should be a narrow bright scattered component .
the only possible sign that we have of this is the polarized off - nuclear emission region in 3c 357 , and the fact that it is close to the radio axis lends weight to this interpretation .
however , the s / n is low in this case , and better observations are needed before this possibility can be discussed . in their discussion of cygnus
a , o97 noted that the larger and more highly polarized fan is on the west side , and that this is the front side of the galaxy , as determined by radio observations .
they commented that this is consistent with increased forward scattering on dust particles .
we do not have the conclusive information provided by vlbi for the other nlrg with fans , but still can compare the size and strength of the fans with other indicators .
we have 3 potential indicators of sidedness : radio structure , polarization fans , and velocity .
table 6 shows the results .
column 2 gives the side with the larger and stronger polarization fan .
as can be seen in figure 5 , these are the regions which have extensions in @xmath220 , although the peak values of @xmath220 are on the opposite side .
we have individual spectra for the two sides and the blue - shifted side is noted ; this side is in front if the material is out - flowing .
we also attempted to establish the side in front with a radio image .
in cygnus a this is unambiguously determined by vlbi . in the others we chose the side that has the more complete and stronger jet , but in 3c 33 we were unable to choose .
it appears in table 6 that the correlations are no more than would be given by chance , and that the cygnus a result can not be generalized .
we have sensitive keck spectral and imaging polarimetry on 13 fr ii radio galaxies five blrg and eight nlrg .
one nlrg is unpolarized and the others have continuum polarizations ranging from 0.7% to 9.6% at 5500 ( rest frame ) .
when @xmath15 is corrected for galactic starlight the polarizations rise to 15% , and in 3c 33 a further correction for fc2 yields a value @xmath270 for the intrinsic polarization of the scattered nuclear light .
high values like this are expected from scattering in an optically thin cone .
broad h@xmath0 is visible in six of the seven polarized nlrg . in five of them
the broad lines can be directly seen in the spectrum of total flux , but they are more prominent in polarized flux . in all the polarized nlrg except 3c 357 , and perhaps 3c 135 , the polarization of broad h@xmath0 after correction for galactic starlight is higher than in the neighboring continuum .
this is evidence for an additional polarization - diluting continuum component , called fc2 .
its nature is unknown but it could be due to hot stars or brehmsstrahlung .
@xmath2 ( measured at the nucleus ) is within 24of perpendicular to the radio axis for all the nlrg ; the mean value of @xmath271
. the blrg are not close to parallel . the median value of @xmath272 , and only one ( 3c 382 ) has @xmath273 .
the b band nlrg polarization images show a symmetric polarization pattern characteristic of a bipolar reflection nebula .
these objects contain a blr and a continuum source which is hidden from us by dust but revealed by the polarization spectra and images . from directions within the scattering cone
they would look like blrg or quasars .
hence , the obscuring torus paradigm unifying seyfert 1 and 2 galaxies also applies to fr ii radio galaxies .
a large fraction of polarized nlrg contain a blr , but we do not know if the correlation between @xmath15 and the blr is intrinsic or merely due to the existence of appropriately located scattering material that produces the polarization and also allows the blr to be recognized .
conversely , we have one unpolarized nlrg that does not show broad lines .
we do not know if the blr is missing or if there is just a lack of scattering material .
we are unable to generalize these results to find , for example , the fraction of nlrg that are polarized , as our sample is strongly biased towards objects known in advance to be polarized .
there is a hint in the low s / n data for 3c 357 that there is no blr , because the west emission region ( w ) appears to partly consist of scattered nuclear continuum light , but there is no broad h@xmath0 .
however , this suggestion is not correct if w is scattering a beamed synchrotron component .
there probably is a variation in the luminosity of the blr in these rg , and that , together with the aspect and extinction , can explain the gradation in broad h@xmath0 seen in figures 1 and 3 .
we follow @xcite in defining an nlrg as one in which the blr ( and continuum source ) is mainly seen by scattering , and a blrg as one dominated by direct nuclear light . by this definition ,
3c 234 is an nlrg although broad lines are easily visible in total flux .
we suggest that trying to draw careful distinctions here is unimportant , and that it is the geometric and physical properties in the nucleus that are of interest .
we propose a model for the blrg in which the spectra are composed of differing fractions of direct and scattered light , with reddening .
the model explains the general blue shape of the polarization spectra , but does not fit 3c 109 and fsc 2217 + 259 very well .
these are the two red blrg with large extinction on the direct ray .
3c 227 and 3c 445 have marked rotations in @xmath2 in the broad lines .
we interpret this to mean that there are multiple lines of sight to the blr .
we have a partial view deep into the nucleus but probably near the edge of the torus .
we suggest that the broad - line clouds are in streaming , not chaotic motion , and we describe a simple scenario of clouds in equatorial orbits that can explain much of the observed rotation phenomena in 3c 445 .
3c 382 was very bright and was a quasar at our observing epochs .
at other epochs it has been several magnitudes weaker and had an easily - visible host galaxy . in this case
the names quasar and blrg have little physical distinction other than to define whether or not the nuclear source is brighter than the stars .
we are grateful to j. beverly oke and judy cohen , first for building the lris and second for the many discussions we had about its use .
we are grateful to the instrument specialists at keck , tom bida and randy campbell , for their support that often went well beyond the call of duty .
we thank i. browne and m. marcha for telling us about the high polarization of fsc 2217 + 259 in advance of publication , and r. fosbury and j. vernet for suggestions concerning dust and other matters .
we thank r. antonucci , l. armus , j. baker , l. ferrarese , d. hines , p. mccarthy and s. di serego alighieri for useful discussions .
we are grateful to r. fosbury and s. di serego alighieri for assistance with the observations , and thank the referee , bev wills , for many useful comments and suggestions .
we used the vital nasa extragalactic data base ( ned ) in this research .
keck observatory is a scientific partnership between the university of california and caltech , made possible by the generous gift of the w.m .
keck foundation and support of its president , howard keck .
this work was partly supported by nsf , grant ast-9121889 , and research by hdt at llnl was supported by doe under contract w7405-eng-48 .
fosbury , r.a.e . ,
vernet , j. , villar - martin , m. , cohen , m. , cimatti , a. , di serego alighieri , s. 1999 , in eso conference on chemical evolution from zero to high redshift " , eso astrophysics symposia , eds j. walsh and m. rosa ( springer ) , in press tadhunter , c.n . ,
dickson , r.d . ,
morganti , r. , & villar - martin , m. 1997 , in
quasar hosts " , eso / iac conference held on tenerife ,
september 1996 , eds .
clements and i. prez - fournon , ( springer ) p. 311 .
llccccl 0106@xmath274130 & 3c 33 & n & 0.0592 & @xmath27549.3 & @xmath2760.3 & 1 , 2 , 18 0404@xmath274035 & 3c 105 & n & 0.0890 & @xmath27533.6 & 1.3 & 0410@xmath274110 & 3c 109 & b & 0.3056 & @xmath27527.8 & 2.3 & 3 , 4 , 5 , 6 0511@xmath274008 & 3c 135 & n & 0.1274 & @xmath27521.0 & 1.1 & 0806@xmath275103 & 3c 195 & n & 0.1100 & 12.0 & 1.1 & 7 , 8 0945@xmath274076 & 3c 227 & b & 0.0862 & 42.3 & @xmath2760.3 & 1 , 3 , 8 , 9 0958@xmath274290 & 3c 234 & n & 0.1848 & 52.7 & @xmath2760.3 & 1 , 3 , 4 , 6 , 7 , 10 , 18 1529@xmath274242 & 3c 321 & n & 0.0961 & 53.9 & 0.4 & 2 , 4 , 7 , 11 , 18 1726@xmath274318 & 3c 357 & n & 0.1661 & 30.6 & 0.5 & 1833@xmath274326 & 3c 382 & b & 0.0579 & 17.4 & 0.7 & 1 , 3 , 9 1957@xmath274406 & cyg a & n & 0.0561 & 5.8 & & 12 , 13 , 14 , 15 , 16 , 18 2217@xmath274259 & fsc 2217 + 259 & b & 0.0850 & @xmath27525.4 & 0.8 & 17 2221@xmath275023 & 3c 445 & b & 0.0562 & @xmath27546.7 & 0.4 & 1 , 3 , 6 , 9 3c 33 & 10 96 & 1500 & 62 & 10 96 & 1200 3c 105 & 12 95 & 2880 & 29 & & 3c 109 & 11 90 , 1 91 & 8400 & 90 & 12 95 & 2400 3c 135 & 10 94 ,
12 95&3600 , 5280 & 165 , 0 & 10 96 & 2400 3c 195 & 12 95 & 6720 & 35 & 12 95 , 4 96 & 2400 , 2400 3c 227 & 12 94 & 1920 & 120 & 12 95 & 1440 3c 234 & 10 94 & 4800 & 90 & 12 95 , 4 96 & 2880 , 2400 3c 321 & 5 96 & 3600 & 130 & 4 96 & 1200 3c 357 & 5 96 & 3600 & 95 & 4 96 & 2400 3c 382 & 5 96 & 1920 & 90 & 4 96 & 120 cyg a & 8 94 , 10 96&3600 , 7920 & 35 , 101 & 10 96 & 3600 fsc 2217 + 259 & 7 95 & 1200 & 140 & 12 95 & 240 3c 445 & 7 94 & 3600 & 159 & & lcccccc 3c 33ne & 8 & 2.4 & 3.1 0.4 & 150.5 2.2 & 0.74 & 11.93c 105 & & 8.6 & 0.0 0.3 & & 0.78 & @xmath2762.73c 109 & 4 & 5 & 6.8 0.2 & 170.5 0.6 & & 3c 135 & 8 & 2.2 & 0.7 0.1 & 142.1 4.8 & 0.85 & 4.73c 195 & 4 & 1.9 & 2.3 0.1 & 128.1 0.9 & 0.62 & 6.13c 227 & 4 & 3.7 & 2.3 0.1 & 41.6 0.7 & 0.30 & 3.33c 234 & 4 & 2.2 & 9.6 0.2 & 156.3 0.4 & 0.35 & 14.83c 321 & 6 & 17.2 & 1.1 0.1 & 40.4 2.7 & 0.66 & 3.2 3c 357w & 8 & 2.2 & 1.9 0.3 & 24.0 6.6 & 0.87 & 14.63c 382 & 4 & 5.4 & 1.0 0.0 & 63.0 0.4 & @xmath2760.10 & 1.0cyg a & 4 & 7.7 & 2.3 0.1 & 8.5 1.7 & 0.70 & 7.7fsc 2217 + 259 & 4 & 4.3 & 7.0 0.2 & 43.0 0.5 & 0.28 & 9.73c 445 & 4 & 8 & 2.1 0.1 & 142.0 1.0 & @xmath2760.10 & 2.1 lccc quasar & & & 3c 273 & n & y & y blrg & & & 3c 109 & y & y & 3c 227 & y & y & 3c 303 & & y & y 3c 382 & n & y & fsc 2217 + 259 & y & y & 3c 445 & n & y & nlrg & & & 3c 33 & y & y & 3c 105 & n & n & 3c 135 & y & y & n 3c 184.1 & & n & y 3c 195 & y & y & 3c 219 & & y & y 3c 223 & & y & y 3c 234 & y & y & y 3c 236 & ( n ) & & n 3c 321 & y & y & n 3c 319 & & n & 3c 327 & & & n 3c 357 & y & n & cyg a & y & y & lccccc blrg & & & & & 3c 109 & -0.58 & 144.0 6 & 1 & 170.5 0.6 & 263c 227 & -1.90 & 84.5 7 & 2 & 41.6 0.7 & 433c 382 & -1.10 & 52.0 & 3 & 63.0 0.4 & 11fsc 2217 + 259 & -0.40 & 158.0 0.5 & 4 & 43.0 0.5 & 653c 445 & -1.41 & 173.0 3.5 & 5 & 142.0 1.0 & 31nlrg & & & & & 3c 33 & -2.25 & 19.5 1 & 6 & 91.0 3.1 & 723c 105 & -2.24 & 125.5 & 5 & & 3c 135 & -2.72 & 76.0 3 & 5 & 142.1 4.8 & 663c 195 & -1.46 & 19.5 2.5 & 2 & 128.1 1.4 & 713c 234 & -1.16 & 67.0 5 & 7 & 156.3 0.4 & 893c 321 & -1.78 & 134.5 0.7 & 1 & 61.0 2.0 & 743c 357 & -2.21 & 110.0 & 8 & 24.0 6.6 & 86cyg a & -3.37 & 284.0 2 & 9 & 8.5 1.0 & 84 | we have made high - quality measurements of the polarization spectra of 13 fr ii radio galaxies and taken polarization images for 11 of these with the keck telescopes .
seven of the eight narrow - line radio galaxies ( nlrg ) are polarized , and six of the seven show prominent broad balmer lines in polarized light .
the broad lines are also weakly visible in total flux .
some of the nlrg show bipolar regions with roughly circumferential polarization vectors , revealing a large reflection nebula illuminated by a central source .
our observations powerfully support the hidden quasar hypothesis for some nlrg . according to this hypothesis ,
the continuum and broad lines are blocked by a dusty molecular torus , but can be seen by reflected , hence polarized , light .
classification as nlrg , broad - line radio galaxy ( blrg ) , or quasar therefore depends on orientation . however , not all objects fit into this unification scheme .
our sample is biased towards objects known in advance to be polarized , but the combination of our results with those of hill , goodrich , & depoy ( 1996 ) show that at least 6 out of a complete , volume and flux - limited sample of 9 fr ii nlrg have broad lines , seen either in polarization or p@xmath0 . the blrg in our sample range from 3c 382 , which has a quasar - like spectrum , to the highly - reddened iras source fsc 2217 + 259 .
this reddening sequence suggests a continuous transition from unobscured quasar to reddened blrg to nlrg . apparently the obscuring torus does not have a distinct edge
the blrg have polarization images which are consistent with a point source broadened by seeing and diluted by starlight .
we do not detect extended nebular or scattered emission , perhaps because it is swamped by the nuclear source .
our starlight - corrected blrg spectra can be explained with a two - component model : a quasar viewed through dust , and quasar light scattered by dust .
the direct flux is more reddened than the scattered flux , causing the polarization to rise steeply to the blue .
strong rotations of the electric vector position angle across h@xmath0 in 3c 227 and 3c 445 may be explained by systematic orbital motions in an equatorial broad - line region . |
the top quark ( @xmath0 ) is the heaviest known elementary particle , with a mass that is much higher than those of all the other quarks , and close to the masses of the @xmath1 , @xmath2 and higgs bosons .
its large mass means that it decays quickly without forming hadrons , offering the unique opportunity to study the properties of a ` bare ' quark . with its @xmath3 yukawa coupling to the higgs boson
, it may also be closely connected to electroweak symmetry breaking , or offer a window to physics beyond the standard model ( sm ) .
studies of the properties of the top quark are therefore central to both the lhc and tevatron physics programs .
this review focuses on measurements of the top - pair ( ) production charge asymmetry , and measurements of spin correlation and top polarisation in events , performed by atlas @xcite and cms @xcite in @xmath48tev @xmath5 collisions at the cern large hadron collider .
other properties of the top quark are covered in other reviews presented to the top2014 conference .
one of the most intriguing legacy results from the tevatron is the forward - backward asymmetry .
analyses of the angular distributions of top quarks and anti - quarks by cdf and d0 indicate that the top quarks ( antiquarks ) are produced preferentially following the direction of the proton ( antiproton ) beam .
the latest measurements from cdf suggest that the size of this asymmetry is slightly larger than expected in the sm , whilst the measurements from d0 are in agreement with the sm @xcite .
this asymmetry can not be measured at the lhc , as it collides @xmath5 and not , but an analogous charge asymmetry @xmath6 can be defined by looking at the difference in absolute rapidity values of the produced top quark and antiquark .
the rapidity difference @xmath7 is positive when the top quark is produced at a smaller angle to the beam direction ( large @xmath8 ) than the antiquark , and negative otherwise .
the asymmetry is defined from event counts @xmath9 as : @xmath10 in the sm , this asymmetry is slightly positive for pairs produced via @xmath11 , where interference effects generate a correlation between the direction of the incoming quark and the outgoing top quark , but zero for @xmath12 .
the total resulting asymmetry is an order of magnitude smaller than the tevatron forward - backward asymmetry , _
e.g. _ an nlo qcd calculation including electroweak corrections gives @xmath13 at @xmath4tev @xcite .
the most precise measurements of @xmath6 come from the semileptonic final state , in which the @xmath1 boson from one top quark decays to an electron or muon and a neutrino , and the other to a pair . by selecting events with an isolated electron or muon ,
missing transverse momentum ( ) and at least four jets , atlas and cms both isolate samples of about 60000 events with about 20% non- background , dominated by @xmath1+jets and single top production .
kinematic fits are used to fully reconstruct the system and determine @xmath14 on an event - by - event basis .
the @xmath14 distribution is then unfolded to correct for background , efficiency and resolution effects , and the inclusive @xmath6 corrected back to the parton level extracted .
the measurements from atlas @xcite and cms @xcite at @xmath4tev are shown in table [ t : asym ] , and have been combined @xcite to give a value of @xmath15 , consistent with both zero and the sm prediction .
cms has also measured @xmath6 at @xmath16tev using a very similar analysis @xcite .
.[t : asym ] measurements of the inclusive asymmetry @xmath6 and leptonic asymmetry from atlas @xcite and cms @xcite at @xmath4tev and @xmath16tev with their statistical and systematic uncertainties , together with the lhc combination @xcite and corresponding theoretical predictions @xcite . [ cols="<,<,<,<,<,<",options="header " , ] both collaborations have also measured @xmath6 differentially , as a function of the rapidity , transverse momentum and invariant mass of the system .
the latter in particular increases the sensitivity to new physics scenarios , which are expected to be more prominent at high as shown in figure [ f : acljets ] .
atlas has also measured the asymmetry for @xmath17 , this cut on the longitudinal velocity of the system increasing the fraction of @xmath11 events and the potential new physics contributions , but no significant deviations from the sm have been seen .
measurements of the charge asymmetry @xmath6 as a function of the invariant mass of the system in semileptonic events from atlas at @xmath4tev @xcite and cms at @xmath16tev @xcite .
the measurements are compared to sm nlo qcd predictions with electroweak corrections from ref .
@xcite ( cms nlo1 ) and ref .
@xcite ( cms nlo2 and atlas ) , and to various beyond - standard model scenarios ( see @xcite for details).,title="fig:",width=264 ] measurements of the charge asymmetry @xmath6 as a function of the invariant mass of the system in semileptonic events from atlas at @xmath4tev @xcite and cms at @xmath16tev @xcite .
the measurements are compared to sm nlo qcd predictions with electroweak corrections from ref .
@xcite ( cms nlo1 ) and ref .
@xcite ( cms nlo2 and atlas ) , and to various beyond - standard model scenarios ( see @xcite for details).,title="fig:",width=245 ] the charge asymmetry has also been measured in the dilepton channel at @xmath4tev @xcite . in dileptonic events , the @xmath1 bosons from both top quarks decay to leptons and neutrinos , giving an under - constrained system since the measurement can not resolve the separate contributions of each neutrino .
extra assumptions , such as the expected distribution of neutrino rapidities , are used to find the most probable kinematic configuration for each event , allowing @xmath14 to be extracted and @xmath6 to be measured .
in dileptonic events , a complementary asymmetry observable can be defined , based on the difference in @xmath18 between the positive and negatively charged leptons : @xmath19 .
the distributions of both variables for the updated atlas dilepton analysis @xcite ( new for this conference ) are shown in figure [ f : aclldist ] , and the @xmath6 and values from both atlas and cms are shown in table [ t : asym ]
. unfolded normalised @xmath20 distribution ( left ) and dilepton @xmath21 distribution ( right ) from the atlas dilepton charge asymmetry analysis at @xmath4tev @xcite .
the data is compared to the simulation predictions based on powheg+pythia.,title="fig:",width=230 ] unfolded normalised @xmath20 distribution ( left ) and dilepton @xmath21 distribution ( right ) from the atlas dilepton charge asymmetry analysis at @xmath4tev @xcite .
the data is compared to the simulation predictions based on powheg+pythia.,title="fig:",width=230 ] the inclusive charge asymmetry measurements are summarised in table [ t : asym ] .
the semileptonic measurements have a precision of around 1% , not yet precise enough to distinguish between zero and the small non - zero asymmetry expected in the sm .
the dilepton measurements have lower precision , due to the smaller event samples and ambiguities inherent in the dileptonic final state .
the top quark lifetime of about @xmath22s is much shorter than the time required to form hadrons , so the top quark decays as a ` bare ' quark , transferring information on its spin to the decay products . in pair production ,
the polarisation is expected to be negligible , but the spins of the @xmath0 and @xmath23 are correlated , such that the asymmetry in the numbers of events where the @xmath0 and @xmath23 have like and unlike spins , @xmath24 , is non - zero .
the double - differential cross - section in the decay angles of the two top quarks , and , is given by : @xmath25 where the asymmetry parameter @xmath26 is scaled by the spin - analysing power of the chosen top quark decay products , and and are the cosines of the angles between the top quark decay product and the chosen polarisation axis .
normally the helicity basis is used , in which the polarisation is measured using the direction of the top quark momentum in the rest frame . the spin - analysing power @xmath27 is @xmath28 for positively - charged leptons , @xmath29 for down quarks from the @xmath1 decay , and -0.393 for @xmath30 quarks from the top decay .
measurement of the spin correlations using the above formalism requires the full reconstruction of the system , using techniques similar to those used in the charge asymmetry measurement .
they have been measured at @xmath4tev by cms in dilepton events @xcite ( see figure [ f : cmspol ] ( left ) ) , and by atlas in both dileptonic and semileptonic events @xcite , using both the helicity basis and the so - called maximal basis which is particularly sensitive to correlations in the @xmath12 subprocess .
all measurements are consistent with the spin correlations predicted by the standard model . unfolded distributions of @xmath31 ( left ) and top quark decay angle @xmath32 ( right ) measured by cms in dilepton events at @xmath4tev @xcite , and compared to predictions with and without spin correlations and from the mc@nlo event generator.,title="fig:",width=207 ] unfolded distributions of @xmath31 ( left ) and top quark decay angle @xmath32 ( right ) measured by cms in dilepton events at @xmath4tev @xcite , and compared to predictions with and without spin correlations and from the mc@nlo event generator.,title="fig:",width=207 ] the spin correlation also affects the distribution of , the difference in azimuthal angles ( transverse to the beamline ) of the two leptons in dileptonic events .
both collaborations exploited this variable to demonstrate the existence of spin correlations in @xmath4tev data .
the cms analysis @xcite unfolded the distribution to parton level , correcting for background , acceptance and resolution effects , and compared the result to simulation predictions with and without spin correlations as predicted by the sm ( see figure [ f : spdphi ] ( left ) ) .
the level of spin correlation was quantified from the asymmetry of the parton - level distribution about @xmath33 as @xmath34 , in agreement with the nlo prediction of 0.115 .
atlas @xcite instead compared the detector - level distribution to fully - simulated events with and without spin correlation , quantifying the spin correlation strength with a fit to sm - like correlated and uncorrelated simulation - derived templates .
the fitted fraction of correlated template was measured to be @xmath35 , compatible with unity . at the conference ,
atlas presented a new analysis using the distribution at @xmath16tev @xcite with an optimised dilepton event selection having higher efficiency ; as shown in figure [ f : spdphi ] ( right ) , this is also compatible with the sm expectation , with a fitted @xmath36 , corresponding to a spin correlation strength in the helicity basis of @xmath37 .
the distribution can also be used to set limits on new physics contributions within the selected event sample , for example top squark pair production as shown by atlas @xcite , or an anomalous -gluon interaction parameterised as a chromomagnetic dipole moment as explored by cms @xcite .
measurements of spin correlation using the distribution in dilepton events from cms at @xmath4tev @xcite and atlas at @xmath16tev @xcite , compared to simulation predictions with and without spin correlation at parton level ( cms ) or detector level ( atlas).,title="fig:",width=226 ] measurements of spin correlation using the distribution in dilepton events from cms at @xmath4tev @xcite and atlas at @xmath16tev @xcite , compared to simulation predictions with and without spin correlation at parton level ( cms ) or detector level ( atlas).,title="fig:",width=226 ] the cms collaboration also used the same @xmath4tev dilepton sample to measure the top quark polarisation in events , via the unfolded distribution of the decay angle @xmath32 @xcite .
the results are shown in figure [ f : cmspol ] ( right ) , and are consistent with zero polarisation .
atlas also measured the top polarisation in both dilepton and semileptonic events , looking for polarisation effects which polarise the @xmath0 and @xmath23 quarks with the same ( cp - conserving ) or opposite ( cp - violating ) signs , and again found results compatible with zero @xcite .
the lhc data have already produced a wealth of top quark property measurements , so far mainly at @xmath4tev .
top - pair charge asymmetry measurements are reaching 1% precision , and the inclusion of the full @xmath16tev dataset may allow the small non - zero expected asymmetry to be observed .
no hints of deviations in either the inclusive or differential asymmetries have been seen , strongly constraining various new physics scenarios .
the expected spin correlations have been clearly observed at both @xmath4tev and @xmath16tev , and the top quark polarisation in events has been seen to be consistent with the expectation of zero within about 2% .
many of these measurements are limited by systematics , and advances in modelling or new analysis techniques will be needed to fully exploit the @xmath16tev dataset .
this will be even more important in the upcoming @xmath3814tev run , which in particular will allow top charge asymmetry measurements to be pushed to sub - percent precision and extended to higher invariant masses . | measurements of top quark properties performed at the large hadron collider are reviewed , with a particular emphasis on top - pair charge asymmetries , spin correlations and polarisation measurements performed by the atlas and cms collaborations .
the measurements are generally in good agreement with predictions from next - to - leading - order qcd calculations , and no deviations from standard model expectations have been seen . |
manganites are characterized by a complex interplay of charge , spin , orbital and lattice degrees of freedom .
undoped and weakly doped single layer manganites la@xmath3sr@xmath4mno@xmath2 show antiferromagnetic ( af ) order @xcite , in contrast to three - dimensional ( 3d ) cubic la@xmath3sr@xmath5mno@xmath6 compounds @xcite , where ferromagnetic ( fm ) @xmath7 planes are stacked antiferromagnetically in the undoped case ( one electron per site , @xmath8 ) . as for their 3d counterparts , the properties of the layered systems are strongly influenced by the orbital degrees of freedom of @xmath0 electrons which couple to the spins and thus influence the magnetic order . for @xmath8 , the hopping of @xmath0 electrons is blocked by large coulomb interaction @xmath9 , and charge fluctuations are quenched .
they may be treated by second order perturbation theory which leads to virtual @xmath10 excitations by either @xmath0 or @xmath1 electrons , generating the spin - orbital superexchange @xcite . in the present paper
we intend to focus on the role played by the orbital degrees of freedom of @xmath0 electrons in stabilizing various types of magnetic order in the undoped single layer manganites , and the changes of spin and orbital correlations with increasing temperature .
we study an effective spin - orbital superexchange model derived for the one - dimensional ( 1d ) chain @xcite , generalized here to two - dimensional ( 2d ) planes of layered manganites .
the model depends on spin configuration @xmath11 and takes the form , @xmath12 it consists of : ( @xmath13 ) the superexchange @xmath14 for the core spins formed by @xmath1 electrons , ( @xmath15 ) the orbital superexchange @xmath16 for the @xmath0 electrons , and ( @xmath17 ) a crystal field term @xmath18 . in the limit of large hund s exchange interaction
, we restrict the @xmath0 electron configurations to the subspace with their spins parallel to the core spins at each site , and thus arrive at spinless fermions with an orbital flavor @xmath19 , standing for @xmath20 and @xmath21 orbitals , respectively .
electrons with antiparallel spins are treated in second order perturbation theory in the same way as double occupancies .
we start the discussion with the @xmath1 core spin superexchange term @xmath14 . in reality
it depends on the total number of @xmath22 electrons occupying two interacting mn ions @xcite .
however , in the undoped systems one finds @xmath23 configuration at each site and one may replace this part of the superexchange by a heisenberg hamiltonian with a properly chosen exchange constant @xmath24 favoring af order : @xmath25 for convenience we use classical core spins @xmath26 of unit length , and compensate their physical value @xmath27 by a proper increase of @xmath28 . for classical spins
@xmath26 parameterized by polar angles @xmath29 , the spin product is given by @xmath30 , where the spin orientation enters via @xmath31 depending on the angle @xmath32 between the spins and on the phase @xmath33 , which cancels at @xmath8 .
the superexchange term @xmath16 describes @xmath0 orbital interactions , as derived recently for a 1d chain @xcite from the full coupled spin - orbital dynamics by replacing the spin scalar products by the actual averages @xmath34 .
the orbital superexchange @xmath16 depends then on spin correlations via the factors @xmath35 in eq .
( [ uij ] ) , @xmath36 here , it is most convenient to use directional orbitals along the @xmath37 and @xmath38 axes , with @xmath39 .
the density operators : @xmath40 and @xmath41 , where @xmath42 , are restricted to single occupancies , and @xmath43 also depend on the direction of the bond @xmath44 .
the pseudospin @xmath45 orbital operators are given by @xmath46 finally , the superexchange constant @xmath47 is given by the high - spin excitation energy @xmath48 , see ref .
the first term of eq.([hj ] ) favors fm spin and alternating orbital ( ao ) order , while the second one is optimized by af spin and ferro orbital ( fo ) order both have similar strength and therefore closely compete .
the last term of the hamiltonian is a crystal field splitting of @xmath0 orbitals , @xmath49 with the orbital splitting @xmath50 .
we consider only @xmath51 with out - of - plane @xmath52 orbitals favored , which is suggested by the elongated octahedra in lasrmno@xmath2 along @xmath53 axis @xcite . in order to treat this model
, we employ a markov chain monte carlo ( mcmc ) algorithm for the classical core spins ( see , e.g. ref .
@xcite ) combined with exact diagonalization for the electron degrees of freedom @xcite . in this method , the many - particle hamiltonian for each spin configuration @xmath11 determined by @xmath35 is solved by exact diagonalization .
the lowest states are then used to evaluate the trace over the fermionic degrees of freedom , @xmath54 , which gives the statistical weight for a given spin configuration @xmath11 , and which is sampled by the mcmc .
autocorrelation analysis was used in order to verify that enough configurations were skipped between measurements .
the boltzmann factors of the lanczos eigenstates are measured in order to ensure that only negligible weight is lost when the weight is calculated from the lowest eigenstates and we found that at most one or two percent are missed for @xmath55 , where @xmath56 is inverse temperature .
spin correlations @xmath57 for the undoped system along the @xmath58-direction are gradually weakened when temperature increases ( fig .
[ fig : ss_0h ] ) . by increasing @xmath28 one identifies two distinctly different cases : for @xmath59 the system is fm at low temperature @xmath60 [ fig .
[ fig : ss_j0_0h ] ] , while af spin correlations arise for @xmath61 and @xmath62 , leading to zig - zag lines in figs .
[ fig : ss_j0.02_0h ] and [ fig : ss_j0.05_0h ] .
already a small value @xmath61 suffices to switch the spin order from fm to af , because fm and af terms strongly compete in eq .
( [ hj ] ) .
for the same reason , increasing temperature can easily destroy the magnetic order and both fm and af correlations decrease rapidly around @xmath63 , except for the stronger af superexchange @xmath62 , with more robust af correlations .
next we investigate orbital correlations @xmath64 shown in fig .
[ fig : oo_0h ] , which exhibit complementary behavior to the spin sector .
here we selected @xmath65 operators which correspond to the expected ao order with @xmath66 orbital states in the 2d @xmath7 plane .
one finds that the orbital correlations alternate for the fm spin order , while this alternation is suppressed for increasing @xmath28 , as almost only orbitals within the plane are then occupied ( fo order ) which promotes the af spin order . for @xmath59 , see fig .
[ fig : oo_j0_0h ] , one notes that the ao order remains robust with rising temperature although the spin correlations vanish , see fig . [ fig : ss_j0_0h ] .
remarkably , in the af case one observes the gradual development of ao order as the af spin correlations become weaker . for @xmath61
this crossover happens at fairly low temperature @xmath67 , while it moves to higher temperature for @xmath68 .
as the magnetic order is so closely related to the orbital order in @xmath7 planes , one expects that changing orbital occupation will modify the magnetic order .
therefore , we investigated the electron density within the out - of - plane @xmath52 orbitals , @xmath69 . indeed , @xmath70 in the fm phase at low temperature ( @xmath59 ) , consistent with ao order , while fo order with @xmath71 orbitals ( @xmath72 ) takes over for the strongly af case ( @xmath68 ) .
with rising temperature , @xmath52 orbitals are slightly depleted in the ferromagnet as orbital alternation becomes weaker . on the contrary , their occupation grows in the af case , when such alternation develops .
this happens quite soon for @xmath73 , but takes much longer for @xmath68 , where af spin order also persists to higher temperature , see fig .
[ fig : oo_j0.05_0h ] .
further , if one adds a crystal field splitting @xmath74 in the af phase with @xmath68 , this forces almost all electrons into the out - of - plane @xmath52 orbitals , the situation encountered in lasrmno@xmath2 . with rising temperature , however , electrons redistribute and in - plane @xmath71 orbitals gradually fill in , see fig . [
fig : n_beta ] , which was also observed in experiment @xcite . the average af and fm terms in @xmath0 superexchange can be derived from ref .
@xcite and are given by : @xmath75 , and @xmath76 .
although @xmath77 becomes slightly smaller at higher temperature in the fm phase at @xmath59 [ fig .
[ fig : n_se_beta](b ) ] , the changes are not very marked , because ao order persists .
the fm coupling is therefore still almost the same as at low temperature , even if the magnetic order is lost because of thermal fluctuations . in the af systems ( @xmath61 , @xmath78 ) ,
the fm terms are first quenched by fo order , but become more important with rising temperature while the af ones are reduced and , as before , this happens much sooner for @xmath61 than for @xmath62 .
interestingly , the af @xmath0 couplings are then changed into fm ones with rising temperature [ fig .
[ fig : n_se_beta](c ) ] . summarizing , at finite crystal field @xmath79 , as in lasrmno@xmath2
, one observes that both @xmath77 and @xmath80 are weaker than for @xmath81 .
the reason is that pairs of alternating orbitals are missing for fo order with @xmath52 orbitals in @xmath7 plane ( small @xmath77 ) , and the overlap between @xmath52 orbitals is much smaller than between @xmath71 orbitals ( small @xmath80 ) .
therefore , the af order observed in lasrmno@xmath2 is promoted mainly by @xmath28 . | we study a two - dimensional effective orbital superexchange model derived for strongly correlated @xmath0 electrons coupled to @xmath1 core spins in layered manganites .
one finds that the ferromagnetic ( fm ) and antiferromagnetic ( af ) correlations closely compete , and small changes of parameters can switch the type of magnetic order . for the same reason , spin order is easily destroyed with rising temperature , while alternating orbital correlations can persist to temperatures where fm order has already melted . a scenario for the af phase observed in lasrmno@xmath2
is presented
. published in phys .
stat .
sol .
( b ) * 243 * , 277 ( 2006 ) |
the discovery of turbo codes and the re - discovery of low - density parity - check ( ldpc ) codes have , over the night , closed the theory - practice gap of the shannon capacity limit on additive white gaussian noise ( awgn ) channels .
they have also revolutionized the coding research with a new paradigm of _ iterative probabilistic inference _ , commonly dubbed the _ soft - iterative _ paradigm . since their success with turbo codes and ldpc codes ,
the soft - iterative paradigm has become a vital tool in widespread applications in communication and signal processing . a complex communication system comprised of layers of functional blocks
that were previously individually or sequentially tackled , can now use a `` soft - iterative '' treatment , close in spirit to that of the turbo or ldpc decoder , to achieve quality performance with manageable complexity .
celebrated applications include , for example , iterative demodulation and decoding , turbo equalization ( also known as iterative decoding and equalization ) , and multi - user detection , and iterative sensing and decision fusion for sensor networks .
the significance and wide popularity of the soft - iterative algorithm has caused a considerable amount of study on its behavior , performance and convergence .
present models and methodologies for analyzing estimation and decoding methods may be roughly grouped into the following categories : 1 ) . the most straight - forward way to evaluate the performance of an estimation / detection / decoding method ,
be it soft - iterative or otherwise , is through monte carlo simulations .
the result is very accurate , but the simulation is usually lengthy , tedious , and not scaling well .
additionally , simulations do not shed much insight into why the performance is so and how the performance might be improved .
classic analytical methods come from the perspectives of maximum likelihood ( ml ) , maximum _ a posteriori _ ( map ) probability , or minimum square error ( mse ) .
they inevitably assume that the subject method is optimal and always deciding on the candidate that has the largest probability , maximum likelihood ratio , or the minimal hamming / euclidean distance to what s been observed .
they produce useful performance bounds , but may present a non - negligible gap to the true performance of the practical , iterative estimator at hand .
3 ) . powerful iterative analytical methods , notably the _ density evolution _ ( de ) @xcite and the _ extrinsic information transfer _ ( exit ) charts @xcite , were developed in the last decade .
these methods faithfully capture the iterative trajectory of many estimators / decoders , and have unveiled several fundamental and intriguing properties of the system ( e.g. the `` convergence property '' and the `` area property '' ) @xcite . however , several underlying assumptions thereof , including the ergodicity assumption , the neighborhood independence assumption and the gausianity assumption , make them suitable mostly for evaluating the _ asymptotic _ behavior ( i.e. infinite block size ) .
many real systems have limited lengths of a few hundred to a few thousand ( bits ) , and the accuracy and usefulness of these methods can become limited in such cases . 4 ) .
to tackle the hard problem of iterative analysis for short - length signal sequence , researchers have also developed several interesting concepts and ideas , including _ pseudo codewords _
@xcite , _ stopping sets _ and _ trapping sets _
they boast some of the most accurate performance predictions at short lengths .
the drawback , however , is that efficient and systematic ways to identify and quantify these metrics are not readily existent , and hence , one may have to rely on computer - aided search of some type , causing daunting complexity .
the majority of the existing methods , as summarized above , largely stem from a statistical and/or information theoretical root .
they have significantly advanced the field , but are also confronted with challenges and limitations , as they try to use statistical metrics and tools that are based on ensemble averages ( such as mean , variance , entropy , mutual information ) to predict and control the iterative process of a large - dimension , highly - dynamical , and apparently - random signal sequence . the notion that iterative probabilistic inference algorithms can be viewed as complex dynamical systems @xcite-@xcite presents an interesting departure from the existing ensemble - average based methods , and brings up new ways of evaluating the soft - iterative algorithm on individual blocks .
generally speaking , the performance of a system should be assessed in the _ average _ sense , such as the bit error rate ( ber ) averaged over hundreds of thousands of blocks . at the same time , however , it also makes sense to evaluate the _ per - block _ performance , namely , the number ( or the percentage ) of errors in individual blocks . per - block error rate reflects error bursts and/or the worst - case situation , and can be of interest in several applications .
for example , in multimedia transmission , a modest number of bit blip errors in an image may cause only a minor quality degradation that is hard to perceive by human eyes , whereas excessive errors will cause the image to be badly distorted and unusable . in magnetic recording systems , a reed solomon ( rs ) outer code is generally employed after the channel - coded partial - response ( pr ) channel , to clear up the residual errors left by the equalizer / decoder .
it does not really matter that every block has errors ( after the equalizer / decoder ) ; as long as the per - block error rate is within the error correction capability of the rs wrap , zero - error is achievable for the entire system .
hence , the specific issue we investigate here is how the per - block performance improves with the number of iterations .
common wisdom has it that more iterations can not hurt , i.e. , a larger number of iterations may not necessarily lead to ( worthy or noticeable ) performance gains , but it can not degrade the performance either .
this apparent truth , as verified by the numerous studies reported on the bit error rate ( ber ) simulations , density evolution ( de ) analysis , and extrinsic information transfer ( exit ) charts , holds in the context of _ average _ performance . in terms of the per - block performance , however , our studies reveal that it is not only possible , but also quite likely , for an individual block to encounter an `` fluctuating '' decoding state , such that the number of errors in that block keeps bouncing up and down ( rather than monotonically decreasing ) with the iterations .
what this phenomenon , thereafter referred to as the _ z - crease _ phenomenon , implies in practice is that a larger number of iterations are not always beneficial , and that the right timing may play a more important role .
if the decoder stops at an unlucky iteration , it may actually generate far more errors than if it stopped several iterations earlier .
it is worth noting that z - crease is not special ; it is actually a _
universal _ phenomenon that vastly exists in iterative decoding and estimation systems .
we have examined a variety of different systems , including low - density parity - check ( ldpc ) codes with message - passing decoding , turbo codes with turbo decoding , product accumulate ( pa ) codes with iterative pa decoding @xcite , and convolutionally - coded inter - symbol interference ( isi ) channel with turbo equalization . in all of these systems ,
we have observed the z - crease phenomenon .
to study the per - block system behavior and reveal the z - crease phenomenon , our approach is to treat the iterative estimation / decoding system a high - dimensional nonlinear dynamical system parameterized by a set of parameters , to further transform it to a one - dimensional state space with a single parameter , and examine the time evolution of the states at specific signal - to - noise ratios ( snr ) . in this paper , we take a popular turbo code as an example , and report here the observation of a wide range of phenomena characteristic to nonlinear dynamical systems , including several new phenomena not reported previously .
a communication or signal processing system generally consists of many processing blocks inter - connected in parallel , in serial , or in hybrid , each fulfilling a specific task .
since the solution space of an `` integrated '' process is the kronecker product of all the constituent solution spaces , to launch an overall optimal solution usually induces prohibitive complexity .
a more feasible solution is to apply iterative algorithms , which allow constituent sub - units to perform local process and to iterative exchange and refine processed `` messages '' , thus achieving a solution considerably better than that from sequential processing with a manageable complexity .
iterative algorithms are by nature probabilistic inference based , where the `` messages '' to be processed and communicated represent the reliability or confidence level of a digital decision , commonly formulated as _ log - likelihood ratios ( llr ) _ ; but they can also be modeled as ( nonlinear ) dynamical systems . to help model all the variants of iterative algorithms in a universal mathematical formulation ,
we have summarized some the properties assumed to be features of these algorithms : ( i ) an iterative algorithm is a dynamical system with a large number of dimensions , possibly depending on many parameters , and distances along trajectories increase ( decrease ) polynomially , sub - polynomially or exponentially .
( ii ) it is formed by two or more units interacting with each other ; each unit responds to messages received from the others in a nonlinear manner .
( iii ) the system is in general hierarchical : a message may be treated in several different levels ( units ) before reaching the center of action .
( iv ) the system in its evolution may be adaptive , i.e. with memory .
( v ) local interactions may have global effect : they may produce considerable global change in the system over the time , e.g. the `` wave effect '' in the decoding of an irregular ldpc code .
we start by evaluating the turbo decoder , which is useful in its own right , and whose information theoretical analysis has reached a good level of maturity . a typical rate-1/3 turbo code , depicted in fig .
[ figure:1turbo ] , is formed of two constituent recursive systematic convolutional ( rsc ) codes , concatenated in parallel through a pseudo - random interleaver .
it encodes a block of @xmath0 binary bits @xmath1 into a codeword of @xmath2 binary bits , @xmath3 $ ] .
the decoder operates much like the turbo engine in an automobile , in which two sub - decoders perform soft - in soft - out decoding , and iteratively exchange and refine llr messages @xmath4 corresponding to @xmath1 .
let @xmath5 $ ] be the noise , induced by the physical channel or transmitter / receiver circuitry , and let @xmath6 be the noisy observation available at the decoder . exploiting the geometric uniformity of the codeword space of a turbo code ( or any practical ecc ) , we can model the turbo decoder as a discrete - time dynamical system in constant evolution : @xmath7 where the superscript @xmath8 denotes the number of half iterations , @xmath9 , @xmath10 and @xmath11 are the parameters of the dynamical system , and @xmath12 and @xmath13 are nonlinear functions describing the constituent rsc sub - decoders , reflecting in general an implementation of the bahl - cocke - jelinek - raviv ( bcjr ) decoding algorithm , the soft viterbi algorithm ( sova ) , or their variations . when @xmath0 takes on a value of a few thousand or larger , as in a practical scenario , this @xmath0-dimensional @xmath2-parametrized nonlinear dynamical system becomes too complex to characterize or visualize . to make the problem tractable
, we propose to `` project '' these dimensions to one or a few `` critical '' ones .
borrowing insight developed from conventional decoder analysis and after performing a careful evaluation , we propose to project the @xmath2 parameters into a single parameter , the ( approximated ) signal - to - noise ratio ( snr ) , @xmath14\ , ||^2 $ ] , ( where @xmath15 is the code rate ) , and to project the @xmath0 dimensions of the state space into one dimension , the mean magnitude of llrs , @xmath16 $ ] where @xmath17 .
the nonlinear dynamical system remains in constant evolution in the @xmath0-dimensional space with @xmath2 parameters ( as a real turbo decoder does ) , but characterizing the system using reduced dimensions drastically simplifies the analysis , enabling a better visualization and understanding of the further behavior .
[ figure:1turbo ]
consider noise samples represented by @xmath5 = [ z_1 , z_2 , . .
. , z_{3k}]$ ] .
different vectors of noise samples are said to have the same noise realization , if they have the same fixed ratios between consecutive sample values , @xmath18 , @xmath19 , @xmath20 , @xmath21 .
thus for a given noise realization , the noise vector @xmath22 is completely determined by the ( approximated ) snr @xmath23 . in general , @xmath0 should be chosen sufficiently large to make @xmath23 a close approximation of the true channel snr .
extensive simulations are performed in our preliminary study . a whole range of phenomena known to occur in nonlinear dynamical systems , including fixed points , bifurcations , oscillatory behavior ,
period - doubling , limited cycles , chaos and transient chaos , are observed in the iterative decoding process as @xmath23 increases ( fig .
[ fig:2a][fig:2i ] ) .
some of these phenomena were noted in previous studies @xcite @xcite , but we report interesting new discoveries . for each motion type ,
we provide two pictures : a wave picture illustrating the change of mean magnitude of llrs @xmath24 and the minimum magnitude of llrs @xmath25 ( y - axis ) as a function of the number of half iterations ( x - axis ) , and a trajectory picture presenting the phase trajectory from one half iteration to the next . in some cases ,
we also present a third picture of a zoomed - in trajectory after 500 half iterations db , indecisive fixed point .
( left : wave picture ; right : trajectory picture.),title="fig:",width=172 ] db , indecisive fixed point .
( left : wave picture ; right : trajectory picture.),title="fig:",width=172 ] the iterative process inevitably starts with and ends at a fixed point .
the former , occurring at an asymptotically low snr ( such as @xmath26 db in our experiment , fig .
[ fig:2a ] ) , is termed an _ indecisive fixed point _ , and is associated with an unacceptably high error probability ( @xmath27 in our experiment ) . the latter ,
occurring at an asymptotically high snr ( e.g. , @xmath28 db , fig . [ fig:2i ] ) , denotes a successful decoding convergence to a zero - error _ unequivocal fixed point_. db , periodical fixed point losing stability.,title="fig:",width=172 ] db , periodical fixed point losing stability.,title="fig:",width=172 ] [ fig:2b ] db , periodical fixed point losing stability.,title="fig:",width=172 ] db , periodical fixed point losing stability.,title="fig:",width=172 ] db , periodical fixed point losing stability.,width=172 ] db , periodical fixed point losing stability.,title="fig:",width=172 ] db , periodical fixed point losing stability.,title="fig:",width=172 ] db , periodical fixed point losing stability.,width=172 ] db , chaos.,title="fig:",width=172 ] db , chaos.,title="fig:",width=172 ] db , chaos.,title="fig:",width=172 ] + db , chaos.,title="fig:",width=172 ] db , chaos.,title="fig:",width=172 ] db , chaos.,width=172 ] db , chaos.,title="fig:",width=172 ] db , chaos.,title="fig:",width=172 ] between the two asymptotic ends are a myriad spectrum of bifurcations , some of which correspond well to the ostensible concepts and phenomena from the information theory , and others appear foreign and await an indepth study . as @xmath23 increases from @xmath29 to @xmath30 db ( fig .
[ fig:2b ] , the system remains trapped to an indecisive fixed point , but the long convergence time indicates the stability of the indecisive fixed point begins to break down . at snr of @xmath31 db , the indecisive fixed point undergoes a flip bifurcation and a stable _ periodical fixed point _ with period of 5 is formed ( fig . [ fig:2c ] ) .
further increasing @xmath23 to @xmath32 db leads to an increased period from 5 to 10 , showing the _ period - doubling _ phenomenon ( fig .
[ fig:2d ] ) .
a closer inspection , shown in the zoomed - in trajectory picture , indicates that the motion is not exactly repetitive , but follows an approximate periodic orbit . as been verified in the experiment here ( as well as other complex systems )
, the period will continue to double without bound as @xmath23 increases .
the `` discrete '' orbit eventually becomes continuous at @xmath33 db , presenting a _ limited cycle _ or _ limited ring _ a closed curve homeomorphic to a circle ( fig .
[ fig:2e ] ) . as snr further increases , the limited ring loses its stability , and converges ones again to an _ indecisive fixed point _ at @xmath34 db ( fig .
[ fig:2f ] ) .
this is rather surprising and is the first time that this type of indecisive fixed points has been observed for turbo decoders .
unlike the fixed points at both asymptotic ends , here the phase trajectory oscillates with diminishing amplitude and it takes longer to converge .
db , periodic fixed point.,title="fig:",width=172 ] db , periodic fixed point.,title="fig:",width=172 ] db , periodic fixed point.,width=172 ] db , periodic fixed point.,title="fig:",width=172 ] db , periodic fixed point.,title="fig:",width=172 ] db , periodic fixed point.,title="fig:",width=172 ] db , periodic fixed point.,title="fig:",width=172 ] db , periodic fixed point.,width=172 ] db , unequivocal fixed point.,title="fig:",width=172 ] db , unequivocal fixed point.,title="fig:",width=172 ] db , unequivocal fixed point.,title="fig:",width=172 ] db , unequivocal fixed point.,title="fig:",width=172 ] next at @xmath35 db , the fixed point undergoes neimark - sacker bifurcation through which the phase trajectory goes into an invariant set and , after a transient period , becomes _ chaos _ ( fig .
[ fig:2 g ] ) .
the previous limited studies have suggested that after chaos will be transient chaos and then the convergence to an unequivocal fixed point @xcite .
it is intriguing indeed to report there actually exist a rich variety of motion types between chaos and the asymptotic unequivocal fixed point .
they include : a _ quasi - periodic fixed point _ at @xmath36 db ( fig .
[ fig:2h ] ) , _ transient chaos _ at @xmath37 db ( fig .
[ fig:2i ] ) , a _ periodic fixed point _ at @xmath38 db ( fig .
[ fig:2j ] ) , another _ transient chaos _ with a short transient lifetime at @xmath39 db ( fig . [ fig:2k ] ) , and eventually the zero - error _ unequivocal fixed point _
[ fig:2l ] ) .
repeated tests on a large sample of random noise realizations show that although different realizations produce different bifurcation diagrams , the entire snr range nonexclusively falls apart into three regions : a low - snr region corresponding to stable indecisive fixed points , a transition region known in the communication jargon as the _ waterfall region _ in which bifurcations occur , and a high - snr region corresponding to stable unequivocal fixed points .
it can be proven that ( 1 ) for an iterative estimator / decoder that is probabilistic inference based , given any noise realization @xmath40 and positive number @xmath41 , there exists an snr threshold @xmath42 , such that for any snr @xmath43 , the iterative algorithm converges , with a probability greater than @xmath44 , to a unique and stable indecisive fixed point ; ( 2 ) likewise , there exists an snr threshold @xmath45 , such that for any snr @xmath46 , the iterative estimator / decoder starting with an unbiased initialization converges , with a probability greater than @xmath44 , to a stable unequivocal fixed point that corresponds to zero decoding errors . also of interest
is the rich variety of fixed points we observed in the waterfall region , none of which were reported previously .
these fixed points behave much like chaotic ( sensitive ) non - hyperbolic attractors .
( chaos is a special class of aperiodic , nonlinear dynamical phenomenon , and is characterized by a prominent feature of `` sensitivity to initial conditions . ''
this feature , commonly known as the `` butter - fly effect '' , states that a small perturbation to the initial state would lead to huge and drastically different changes later on . ) in comparison , the fixed points at the two extreme ends of snr are hyperbolic attractors , where distances along trajectories decrease exponentially in complementary dimensions in the ambient space .
these newly observed fixed points , some or all of which may or may not occur depending sensitively on the specific noise realization , are usually associated with a few detection errors .
they appear to provide support , from the dynamical system perspectives , for the information theoretic conjecture that there exists one or more pseudo codewords in the vicinity of a correct codeword ( i.e. a few bits of hamming distance away ) @xcite .
they may also correspond well to the coding concept of _ stopping set _ and _ trapping set _ , which characterize an high - snr undesirable convergence in the bsc ( binary symmetric channel ) decoding model and the gaussian decoding model , respectively @xcite .
it is particularly worth noting that in almost all the motion stages , the mean and the minimum magnitude of llrs fluctuate in a rather notable manner as the number of iteration increases ( see the wave pictures ) . since the mean magnitude of llr
is shown to relate fairly well with the percentage of errors occurred in each frame @xcite , it is therefore reasonable to predict that the per - block error number will also fluctuate with iterations .
for example , for a short block of 1024 bits , we have observed that the number of errors in a particularly block can easily vary between 80 and 120 , in a `` quasi - periodic '' z - crease manner . in other words , it is highly likely that an early lucky iteration may save both complexity and @xmath47 less of erroneous bits than a longer , unlucky iteration . since where the decoder stops
also makes a difference in per - block performance , questions arise as how to stop at the right iteration , and how much benefit there is .
we propose the following rule of thumb to detect z - crease : \i ) the minimum magnitude of llr , @xmath48 , is a very accurate indicator of whether or not the iterative decoder has successfully converged to the unequivocal attractor ( i.e. the correct codeword ) . in a correct convergence ( i.e. when the attractor is the unequivocal fixed point or the unequivocal chaos attractor ) , the minimum magnitude and the mean magnitude of llr will both increases with iterations .
otherwise , the iterative process is trapped in some local minimum ( which corresponds to the indecisive fixed point , the quasi - periodical cycles , and the indecisive chaos ) . in such as case , the average magnitude of llr @xmath49 may continue to increase with iterations at a decent pace , but the minimum magnitude @xmath48 will remain at a very low value , sending a clear signal of unsuccessful convergence .
it is thus convenient to set a threshold for @xmath49 to indicate decoding success .
\ii ) the z - crease is most prominent ( with large error fluctuation ) in the quasi - periodical cycle and the indecisive chaos stages .
hence , it is beneficial to detect z - crease phenomenon as early as possible and to terminate decoding at the earliest `` best '' iteration .
since the z - crease of the bit errors is almost always accompanied with a z - crease of @xmath49 , we suggest using @xmath49 to detect the z - crease of errors . here is a simple but rather effective method : each local maximum point of @xmath50 is taken as a _ candidate point _ , which is like the `` local optimal point '' .
we predict that z - crease is occurring when the @xmath49 value of any one candidate point is lower than the @xmath49 value of its previous candidate point .
following these observations , we also propose a heuristic stopping criterion and suggest performing iterative decoding in the follow manner : the iterative decoder keeps track of @xmath51 and @xmath52 , and terminate when any one of the following conditions happens : 1 .
when @xmath51 increases above a threshold .
2 . when @xmath49 of any one candidate point is lower than @xmath49 of the previous candidate point .
if the decoder stops at condition 1 ) , the current bit decisions and the current iteration are considered as our `` best shots . ''
if the decoder stops at condition 2 ) , then we suggest the decoder trace back to the previous candidate point and use the bit decisions of that iteration as the final decision .
otherwise , the decoder will proceed to reach the maximum iteration cap without voluntary stop .
it should be noted that previous researchers have also used the mean magnitude of llr for early stopping purpose @xcite , but it was used in a different way that did not recognize the z - crease phenomenon . to the best of our knowledge
, the minimum magnitude of llr has not been exploited previously .
our stopping criterion here is most useful in improving the worst - case per - block performance , but not so much for the average performance ( averaged over lots of blocks ) .
it can also cut down the iteration number by @xmath53 or even larger ( especially at low @xmath23s or when the specific frame encounters lucky deep distortion ) , without sacrificing the average performance .
we report the z - crease phenomenon in soft - iterative decoding systems , and use the theory of nonlinear dynamics to justify its existence and generality .
we show that while the average system error rate performance in general improves ( or , does not deteriorate ) with iterations , for individual frames , more iterations may actually do harm to the decoding decisions . analyzing the dynamical behavior of the system , we further propose a simple stopping criterion based on the minimum magnitude and the mean magnitude of llr to detect successful convergence and determine the right iteration to stop .
chung , r. urbanke and t. j. richardson , `` analysis of sum - product decoding of low - density parity - check codes using a gaussian approximation , '' _ ieee trans .
inf . theory _ ,
657 - 670 , feb . 2001 .
l. kocarev , f. lehmann , g. m. maggio , b. scanavino , z. tasev , and a. vardy , `` nonlinear dynamics of iterative decoding systems : analysis and applications , '' _ ieee trans .
inf . theory _ ,
pp . 166 - 1384 ,
april 2006 . | iterative probabilistic inference , popularly dubbed the soft - iterative paradigm , has found great use in a wide range of communication applications , including turbo decoding and turbo equalization .
the classic approach of analyzing the iterative approach inevitably use the statistical and information - theoretical tools that bear ensemble - average flavors .
this paper consider the per - block error rate performance , and analyzes it using nonlinear dynamical theory . by modeling the iterative processor as a nonlinear dynamical system , we report a universal `` z - crease phenomenon : '' the zig - zag or up - and - down fluctuation rather than the monotonic decrease of the per - block errors , as the number of iteration increases .
using the turbo decoder as an example , we also report several interesting motion phenomenons which were not previously reported , and which appear to correspond well with the notion of `` pseudo codewords '' and `` stopping / trapping sets . ''
we further propose a heuristic stopping criterion to control z - crease and identify the best iteration .
our stopping criterion is most useful for controlling the worst - case per - block errors , and helps to significantly reduce the average - iteration numbers . |
a mapping @xmath4 from a complex vector space @xmath5 to a complex vector space @xmath6 is _ semilinear _ if @xmath7 for all @xmath8 and @xmath9 .
we write @xmath10 if @xmath4 is a linear mapping and @xmath11 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } v$ ] ( using a dashed arrow ) if @xmath4 is a semilinear mapping .
we give a canonical form of matrices of a _ cycle _ of linear and semilinear mappings @xmath12^{\mathcal a_t } \ar@{-}[r]^{\ \ \mathcal a_1 } & v_2\ar@{-}[r]^{\mathcal a_2 } & { \ \dots\ } & { v_{t-1 } } \ar@{-}[l]_{\mathcal a_{t-2 } } \ar@{-}[r]^{\ \mathcal a_{t-1}\ \ } & { v_t } } \end{split}\ ] ] in which each line is a full arrow @xmath2 or @xmath3 , or a dashed arrow @xmath13 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } $ ] or @xmath14 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } $ ] .
its special cases are the canonical forms of * matrix pencils , contragredient matrix pencils , and pairs consisting of a linear mapping and a semilinear mapping ( i.e. , cycles @xmath15 \longrightarrow \end{matrix}v_2 $ ] , @xmath15 \longleftarrow \end{matrix}v_2 $ ] , and @xmath15 { \text{\raisebox{0.9 mm } { \ \begin{tikzpicture}[->,thick , xscale=0.56 ] \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } \end{matrix}v_2 $ ] ) ; their canonical matrices were obtained by kronecker @xcite , dobrovol@xmath16skaya and ponomarev @xcite ( see also @xcite ) , and djokovi @xcite ; * cycles of linear mappings ( all arrows in are full ) ; their canonical form is well known in the theory of quiver representations ; see ( * ? ? ?
* section 11.1 ) and @xcite .
the proof is based on the canonical forms of linear and semilinear operators ( i.e. , cycles and ) given by jordan @xcite and haantjes @xcite ( see also theorem [ tej ] ) .
we use the methods of @xcite . in section [ s2 ]
we formulate theorem [ t1.1 ] about a canonical form of matrices of ; in section [ s3 ] we prove it as follows . in section [ ss1 ]
we reduce the proof of theorem [ t1.1 ] to cycles in which @xmath17 and the mappings @xmath18 are linear . in section [ ss2 ] we give a canonical form of the matrices of @xmath19 . in section [ ss3 ]
we give a canonical form of the matrix of @xmath20 with respect to those transformations that preserve the matrices of @xmath19 . in section [ ss4 ]
we complete the proof of theorem [ t1.1 ] .
all matrices and vector spaces that we consider are over the field of complex numbers .
we denote by @xmath21_e$ ] the coordinate vector of @xmath22 in a basis @xmath23 , and by @xmath24 the transition matrix from @xmath23 to a basis @xmath25 . we write
@xmath26 $ ] for a matrix @xmath27 $ ] .
let @xmath11 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } v$ ] be a semilinear mapping .
we say that an @xmath28 matrix @xmath29 is the _ matrix of @xmath30 in bases _ @xmath23 of @xmath5 and @xmath31 of @xmath6 if @xmath32_f=\overline{{\cal a}_{fe}[u]_e}\qquad \text{for all } u\in u.\ ] ] therefore , the columns of @xmath33 are @xmath34_f } , \dots , \overline{[{\cal a}e_n]_f}$ ] .
we write @xmath35 instead of @xmath36 if @xmath37 .
the _ direct sum _ of matrix sequences @xmath38 and @xmath39 is the sequence @xmath40 for each @xmath41 , define the matrices @xmath42 @xmath43 and @xmath44 the following theorem is the main result of the article .
[ t1.1 ] for each system of linear and semilinear mappings , there exist bases of the spaces @xmath45 in which the sequence of matrices of @xmath46 is a direct sum , determined by uniquely up to permutation of summands , of sequences of the following form @xmath47 in which the points denote sequences of identity matrices or @xmath48 : * * * @xmath49 in which @xmath50 , if the number of dashed arrows in is even , * * @xmath49 and @xmath51 in which @xmath52 is real and positive and @xmath53 is either not real or is real and negative , if the number of dashed arrows in is odd ; * @xmath54 with @xmath55 at position @xmath56 ; * @xmath57 in which * * @xmath58 or @xmath59 if @xmath60 and @xmath61 have opposite directions in @xmath62 * * @xmath63 or @xmath64 if @xmath60 and @xmath65 have the same direction in .
note that @xmath66 and @xmath67 in have the size @xmath68 .
it is agreed that there exists exactly one matrix , denoted by @xmath69 , of size @xmath70 and there exists exactly one matrix , denoted by @xmath71 , of size @xmath72 for every nonnegative integer @xmath73 ; they represent the linear mappings @xmath74 and @xmath75 and are considered as zero matrices
. then @xmath76 and @xmath77 for every @xmath78 matrix @xmath79 .
* the composition of a linear mapping @xmath85 and a semilinear mapping
@xmath86 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } w$ ] is the semilinear mapping with matrix @xmath87 * the composition of a semilinear mapping
@xmath88 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } v$ ] and a linear mapping @xmath89 is the semilinear mapping with matrix @xmath90 the identity follows from observing that @xmath91 is a semilinear mapping and @xmath92_{g}= [ { \cal b}({\cal a}u)]_{g}= \overline{{\cal b}_{gf}[{\cal a}u]_{f}}= \overline{({\cal b}_{gf}{\cal a}_{fe})[u]_{e}}\ ] ] for each @xmath93 .
the identity follows from observing that @xmath91 is a semilinear mapping and @xmath92_{g}= [ { \cal b}({\cal a}u)]_{g}= { \cal b}_{gf}[{\cal a}u]_{f}= { \cal b}_{gf}\overline{{\cal a}_{fe}[u]_{e}}= \overline{(\overline{{\cal b}}_{gf}{\cal a}_{fe})[u]_{e}}\ ] ] for each @xmath93 .
let @xmath94 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } v$ ] be a semilinear operator , let @xmath33 be its matrix in bases @xmath95 of @xmath5 and @xmath96 of @xmath6 , and let @xmath97 be its matrix in other bases @xmath98 and @xmath99 .
then @xmath100 since the right hand matrix satisfies with @xmath101 instead of @xmath102 : @xmath103_{e'}}= s_{f\to f'}^{-1}\overline{{\cal a}_{fe } [ v]_{e}}=s_{f\to f'}^{-1}[{\cal a } v]_f=[{\cal a } v]_{f'}\ ] ] in particular , if @xmath37 , then @xmath104 and so @xmath105 and @xmath106 are consimilar ; recall that two matrices @xmath107 and @xmath108 are _ consimilar _ if there exists a nonsingular matrix @xmath109 such that @xmath110 .
it suffices to prove theorem [ t1.1 ] for cycles @xmath30 of the form in which @xmath115 since theorem [ t1.1 ] for the cycles and of length @xmath116 follows from the jordan canonical form and theorem [ tej ] , and all @xmath117 can be made linear mappings by using the following procedure : the semilinear mapping @xmath119 becomes linear in @xmath120 . indeed ,
if @xmath121 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } } \!\
! v_k$ ] , then @xmath122 for all @xmath123 and @xmath124 .
if @xmath125 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } } \!\
! v_k$ ] , then @xmath126 for all @xmath123 and @xmath127 .
deleting @xmath20 in satisfying , we obtain a system of linear mappings @xmath131 in which each line is either @xmath2 or @xmath3 .
the classification of such systems is well known in the theory of quiver representations ( see , for example , gabriel s article @xcite , in which he introduced the notion of quiver representations and described all quivers with a finite number of nonisomorphic indecomposable representations ) .
we recall this classification in lemma [ xtu ] in a form that is used in the next section for reducing the matrix of @xmath132 to canonical form .
a _ chain of type @xmath133 _ of system , @xmath134 , is a sequence of nonzero elements @xmath135 such that @xmath46 act on them as follows : @xmath136 in which the arrows are directed as in ; for simplicity of notation , we write @xmath137 instead of @xmath138 and @xmath139 instead of @xmath140 . * for each system of linear mappings , we can choose bases @xmath141 of the spaces @xmath45 so that @xmath142 consists of disjoint chains .
the sequence of their types @xmath143 in which @xmath144 if @xmath145 , is determined by uniquely up to permutation . *
suppose that the types in are numbered so that for each two pairs @xmath146 and @xmath147 * * if @xmath148 and @xmath149 , then either @xmath150 and @xmath151 , or @xmath152 and @xmath153 ; * * if @xmath154 and @xmath155 , then either @xmath156 and @xmath153 , or @xmath157 and @xmath151 .
+ define nonnegative integers @xmath158 and @xmath159 by as follows : @xmath160 + then the following two conditions are equivalent for a pair of nonsingular matrices @xmath161 and @xmath109 : * * there exists another system of bases @xmath162 of the spaces @xmath45 that consists of disjoint chains with the same sequence of their types such that @xmath161 and @xmath109 are the change of basis matrices from @xmath163 to @xmath66 in @xmath164 and from @xmath165 to @xmath166 in @xmath167 . *
* @xmath161 and @xmath109 are upper block triangular matrices @xmath168 in which every @xmath169 is @xmath170 and every @xmath171 is @xmath172 @xmath47see @xmath173 .
if there is @xmath174 such that @xmath175 in , then @xmath176 the change of basis matrix @xmath161 from @xmath163 to @xmath66 in @xmath164 has the upper block triangular form by the following reason . let @xmath142 contain two chains of basis vectors @xmath178 of types @xmath179 and @xmath180 with @xmath149 .
we consider two cases according to the direction of @xmath181 : * if @xmath182 , then @xmath183 for each @xmath184 , and so we can replace the chain @xmath185 by @xmath186 in @xmath187 .
thus , the basis vector @xmath188 can be replaced by @xmath189 , and so the block @xmath190 in is arbitrary .
due to the condition ( i ) in ( b ) , @xmath151 . *
if @xmath191 , then @xmath192 in which @xmath184 and @xmath193 is the subchain of @xmath194 . hence we can replace the chain @xmath195 by @xmath196 , and so @xmath197 can be replaced by @xmath198 .
therefore , the block @xmath199 in is arbitrary . due to the condition ( i ) in ( b ) , @xmath153 .
analogously , the change of basis matrix @xmath109 from @xmath165 to @xmath166 in @xmath167 has the upper block triangular form by the following reason .
let @xmath142 contain two chains of basis vectors @xmath200 of types @xmath201 and @xmath202 with @xmath203 .
* if @xmath204 , then @xmath205 in which @xmath184 and @xmath206 is the subchain of @xmath194 .
hence , we can replace the chain @xmath195 by @xmath196 , and so the basis vector @xmath207 can be replaced by @xmath208 .
therefore , the block @xmath199 in is arbitrary . due to the condition ( ii ) in ( b ) , @xmath153 . * if @xmath209 , then @xmath210 for each @xmath184 , and so we can replace the chain @xmath185 by @xmath211 .
thus , the basis vector @xmath212 can be replaced by @xmath213 .
therefore , the block @xmath190 in is arbitrary . due to the condition ( ii ) in ( b ) , @xmath151 .
let @xmath174 be such that @xmath175 in , let @xmath214 , and let @xmath215 be two chains .
then @xmath216 for each @xmath184 .
hence the basis vectors @xmath217 and @xmath207 can be simultaneously replaced by @xmath218 and @xmath208 .
therefore , @xmath219 in .
let @xmath30 be a cycle of linear and semilinear mappings satisfying the condition .
we suppose that @xmath221 or @xmath222 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } } \!\!\ !
v_t$ ] ( if @xmath223 or @xmath224 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } } \!\
! v_t$ ] , then we renumber the vector spaces @xmath45 in the reverse order ) . deleting @xmath20 from
, we obtain the chain of linear mappings . by lemma [ xtu ]
, there exists a system of bases @xmath141 of the spaces @xmath45 that consists of disjoint chains .
moreover , the bases @xmath163 and @xmath165 are determined up to replacement by bases @xmath66 and @xmath166 such that the change of basis matrices @xmath161 and @xmath109 have the form . thus , the matrix @xmath225 of the linear or semilinear mapping @xmath20 is reduced by transformations @xmath226 this leads to the following definition .
[ zzz ] by a _ marked block matrix _ we mean a block matrix in which some of square blocks are crossed along their main diagonal by a full or dashed line and each horizontal or vertical strip contains at most one crossed block .
a _ block matrix problem _ is the canonical form problem for marked block matrices with respect to the following admissible transformations : * arbitrary elementary transformations within vertical and horizontal strips such that each crossed block is transformed by similarity transformations if its cross - line is full and by consimilarity transformations if the cross - line is dashed ; * additions of a row multiplied by scalar to a row of a horizontal strip that is located above , and of a column multiplied by scalar to a column of a vertical strip that is located to the right .
we say that @xmath227 and @xmath228 are _ equivalent _ if @xmath227 is reduced to @xmath228 by transformations ( i ) and ( ii ) .
the problem is to find a canonical form of a marked block matrix up to equivalence .
it follows from the form of @xmath161 and @xmath109 in that the canonical form problem for @xmath225 with respect to transformations is the block matrix problem for @xmath225 partitioned into horizontal strips conformally to the partition of @xmath109 and into vertical strips conformally to the partition of @xmath161 ; the condition means that the @xmath229 block of @xmath225 is crossed by a full or dash line .
( note that @xmath225 may contain horizontal strips with no rows and vertical strips with no columns , which follows from . )
let @xmath230 $ ] and @xmath231 $ ] be two block matrices with the same number of horizontal strips , the same number of vertical strips , the same disposition of blocks crossed by full lines , and the same disposition of blocks crossed by dashed lines .
the _ block direct sum _ of @xmath227 and @xmath228 is the block matrix @xmath232\ ] ] with the same disposition of blocks crossed by full lines and the same disposition of blocks crossed by dashed lines .
we say that @xmath227 is _ indecomposable _ if it is not @xmath233-by-@xmath233 and it is not equivalent to a block direct sum of matrices of smaller sizes . *
all blocks of @xmath227 are empty except for one crossed block that is nonsingular , or * @xmath227 is equivalent to a matrix whose entries are only 0 s and 1 s with at most one 1 in each row and each column .
the matrix is partitioned into 2 horizontal and 2 vertical strips .
we extend this partition to the entire @xmath236th horizontal strip and to the entire @xmath237th vertical strip of @xmath227 . using transformations
( ii ) , we make zero all entries over @xmath238 and to the right of @xmath238 in the new strips .
let the new division lines do not pass through crossed blocks . then @xmath238 is a direct summand of @xmath227 .
since @xmath227 is indecomposable , we have that @xmath239 , @xmath240 , and all other blocks are empty , which proves the lemma in this case .
let a new division line pass through a crossed block @xmath241 ( @xmath242 or @xmath243 ) .
then we draw the perpendicular division line such that @xmath241 is partitioned into 4 subblocks with square diagonal blocks ( they are crossed by the full or dash line that crosses @xmath241 ) : @xmath244 if a new division line passes through another crossed block , we repeat this procedure
. denote by @xmath245 the marked block matrix obtained from @xmath227 by deleting the new vertical and horizontal strips that pass through @xmath238 in @xmath79 ( if @xmath227 is @xmath246-by-@xmath73 , then @xmath245 is @xmath247-by-@xmath248 ) .
it is easy to see that all transformations ( i ) and ( ii ) with @xmath245 can be obtained from transformations ( i ) and ( ii ) with @xmath227 .
reasoning by induction , we assume that the lemma holds for @xmath245 .
note that @xmath245 can not satisfy the condition ( a ) since @xmath245 contains a strip with @xmath249 rows or columns and without crossed blocks ; this strip is obtained from the strip of @xmath227 that contains @xmath107 in if @xmath242 or @xmath250 if @xmath243 .
therefore , @xmath245 satisfies the condition ( b ) , and so @xmath227 satisfies ( b ) too .
_ case 2 : @xmath79 is crossed . _
the jordan canonical form for similarity ( if the cross - line is full ) and theorem [ tej ] ( if the cross - line is dashed ) ensure that @xmath79 can be reduced to the form @xmath251 , in which @xmath228 is nonsingular and @xmath252 is a nilpotent jordan matrix . if @xmath228 is nonempty , then we make zero all entries over @xmath228 and to the right of @xmath228 by transformations ( ii ) .
since @xmath227 is indecomposable , @xmath253 and the other blocks of @xmath227 are empty , and so @xmath227 satisfies the condition ( a ) .
thus , we can suppose that @xmath254 .
collecting the jordan blocks of the same size by permutations of rows and the same permutations of columns , we reduce @xmath79 to the form @xmath255 we extend the obtained partition of @xmath79 to the entire @xmath236th horizontal strip and the entire @xmath237th vertical strip of @xmath227 and make zero all entries over and to the right of @xmath256 .
denote by @xmath245 the submatrix of @xmath227 obtained from @xmath227 by deleting the `` new '' horizontal and vertical strips containing @xmath256 .
in particular , @xmath79 converts to @xmath257 the set of crossed blocks of @xmath245 consists of the crossed blocks of @xmath227 ( except for @xmath79 ) and the diagonal blocks @xmath258 of @xmath259 if @xmath79 is crossed by a full line , then @xmath258 are crossed by full lines too . if @xmath79 is crossed by a dash line , then all odd - numbered diagonal blocks @xmath260 of @xmath79 ( they are obtained from the jordan blocks of @xmath79 of odd sizes ) are crossed by dash lines ; the even - numbered diagonal blocks @xmath261 of @xmath79 are crossed by full lines .
denote by @xmath262 the marked block matrix obtained from @xmath245 by arrangement of its `` new '' vertical strips ( passing through @xmath259 ) in reverse order . in particular , @xmath263 converts to @xmath264 * add columns of the @xmath266nd vertical strip to columns of the @xmath267st vertical strip . since @xmath79 is transformed by ( con)similarity transformations , we must do the ( con)inverse transformations with rows of @xmath79 : to subtract the corresponding columns ( or the complex conjugate columns ) of the @xmath267st horizontal strip from the @xmath266nd horizontal strip , which does not change @xmath79 since its @xmath267st horizontal strip is zero .
thus , we can add in @xmath268 columns passing through @xmath269 to columns passing through @xmath270 .
* add columns of the @xmath271th vertical strip to columns of the @xmath267st vertical strip ; the ( con)inverse transformations does not change @xmath79 .
thus , we can add in @xmath268 columns passing through @xmath272 to columns passing through @xmath270 .
* add columns of the @xmath271th vertical strip to columns of the @xmath266nd vertical strip .
the ( con)inverse transformations with rows of @xmath79 spoil the ( 4,3 ) block .
it is restored by additions of columns of the @xmath273th vertical strip ; the ( con)inverse transformations do not change @xmath79 .
thus , we can add in @xmath268 columns passing through @xmath272 to columns passing through @xmath269 .
+ * add rows of the @xmath274rd and @xmath275th horizontal strips to rows of the @xmath267st horizontal strip ; the ( con)inverse transformations with columns do not change @xmath79 .
thus , we can add in @xmath268 rows passing through @xmath269 and @xmath272 to rows passing through @xmath270 .
* add rows of the @xmath275th horizontal strip to rows of the @xmath274rd horizontal strip .
the ( con)inverse transformations with columns of @xmath79 spoil the ( 2,6 ) block .
it is restored by additions of rows of the @xmath273th horizontal strip ; the ( con)inverse transformations do not change @xmath79 .
thus , we can add in @xmath268 rows passing through @xmath272 to rows passing through @xmath269 . +
if @xmath79 is crossed by a full line ( i.e. , @xmath79 is transformed by similarity transformations ) , then we take @xmath276 ( in which @xmath277 is @xmath278 , @xmath279 is @xmath280 , ) and obtain @xmath281 .
thus , @xmath263 can be reduced by transformations @xmath282 and so @xmath283 can be reduced by transformations ( i ) .
assume now that @xmath79 is crossed by a dashed line ; that is , @xmath79 is transformed by consimilarity transformations .
then @xmath284 which is illustrated as follows : @xmath285 thus , @xmath260 are transformed by consimilarity transformations and @xmath261 by similarity transformations .
this proves .
reasoning by induction , we assume that the lemma holds for @xmath262 . the matrix @xmath262 can not satisfy the condition ( a ) of the lemma since @xmath268 is nonempty .
hence , @xmath262 satisfies ( b ) , then @xmath262 satisfies ( b ) too .
let @xmath30 be a system of linear and semilinear mappings whose sequence of matrices of @xmath46 can not be decomposed into a direct sum of systems of matrices of smaller sizes .
by lemma [ xtu ] , there exists a system of bases @xmath141 of the spaces @xmath45 that falls into disjoint chains . by lemma [ kwq ] , the bases @xmath141 can be chosen such that either * the matrices @xmath286 of @xmath287 are the identity and the matrix @xmath225 of @xmath288 is nonsingular , or * if @xmath289 or @xmath290 \draw[dashed ] ( 0,0)--(1,0 ) ; \end{tikzpicture}}\ } } v_l$ ] and @xmath291 , then @xmath292 or @xmath293 ; moreover , if @xmath294 and @xmath295 , then @xmath296 . consider the case ( a ) .
let @xmath297 be the change of basis matrices that preserve @xmath298 , then @xmath299 if the arrow between @xmath300 and @xmath301 is full and @xmath302 if the arrow is dashed .
therefore , @xmath225 is reduced by similarity transformations @xmath303 or consimilarity transformations @xmath304 if the number of dashed arrows in is even or odd , respectively . using the jordan canonical form or theorem [ tej ]
, we obtain a sequence of matrices ( i ) from theorem [ t1.1 ] .
consider the case ( b ) .
let us construct the directed graph , whose set of vertices is the set of basis vectors @xmath142 and there is an arrow from @xmath305 to @xmath22 if and only if @xmath306 for some @xmath307 . in the case
( b ) , this graph is a disjoint union of chains . since the sequence of matrices of @xmath46 can not be decomposed into a direct sum , the graph is connected , and so it is a chain .
for example , a system @xmath308^{\mathcal a_2 } & & v_3\ar@{<--}[dr]^{\mathcal a_3}&\\ v_1\ar@{->}[urr]^{\mathcal a_1 } & & & & & v_4\ar@{->}[dll]_{\mathcal a_4}\\ & v_6\ar@{<--}[ul]_{\mathcal a_6 } & & v_5\ar@{<--}[ll]_{\mathcal a_5}&&\\}\ ] ] may have the chain @xmath309^{\mathcal a_2}&&e_{31 } \ar@{<--}[ddr]^{\mathcal a_3}&\\ & & e_{22 } \ar@{<-}[rr]^{\mathcal a_2}&&e_{32 } \ar@{<--}[ddr]^{\mathcal a_3}&\\ e_{11}\ar@{->}[urr]^{\mathcal a_1 } & & & & & e_{41}\ar@{->}[dll]_{\mathcal a_4}\\ & e_{61}\ar@{<--}[ul]_{\mathcal a_6 } & & e_{51}\ar@{<--}[ll]_{\mathcal a_5}&&e_{42 } \ar@{->}[dll]_{\mathcal a_4}\\ & & & e_{52}\\}\ ] ] its mappings @xmath310 are given by the matrices @xmath311 they form a sequence ( iii ) from theorem [ t1.1 ] with @xmath312 .
it is easy to see that all matrix sequences ( ii ) and ( iii ) from theorem [ t1.1 ] can be obtained analogously .
we have proved that for each cycle of linear and semilinear mappings there exist bases of the spaces @xmath45 in which the sequence of matrices of @xmath46 is a direct sum of sequences of the form ( i)(iii ) ; the uniqueness of this direct sum follows from the krull
schmidt theorem for additive categories ( * ? ? ?
* chapter i , theorem 3.6 ) ( it holds for cycles since they form an additive category in which all idempotents split ) .
nazarova , a.v .
roiter , v.v .
sergeichuk , v.m .
bondarenko , application of modules over a dyad for the classification of finite @xmath236-groups possessing an abelian subgroup of index @xmath236 and of pairs of mutually annihilating operators , j. soviet math .
5 ) ( 1975 ) 636654 . | we give a canonical form of matrices of a cycle of linear or semilinear mapping @xmath0 in which all @xmath1 are complex vector spaces , each line is an arrow @xmath2 or @xmath3 , and each arrow denotes a linear or semilinear mapping .
linear and semilinear mappings , similarity and consimilarity , cycles of mappings , canonical forms , pencils and contragredient pencils 15a04,15a21 |
collective motions of nuclear matter occurring in heavy ion collisions are of great interest since they are expected to provide information about the properties of hot and dense nuclear matter and the underlying equation of state ( eos ) @xcite .
flow effects were predicted by hydrodynamical calculations @xcite and experimentally evidenced at lbl - bevalac @xcite . at beam energies
@xmath2@xmath0mev , the interaction between nuclei is dominated by individual nucleon - nucleon scattering and the repulsive component of the mean field .
this leads to a collective deflection of matter to positive angles in the reaction plane _
i.e. _ in the direction of the projectile remnants ( sideward - flow ) .
conversely , at few tens of @xmath0mev , the interaction is dominated by the attractive mean field , so that nucleons emitted in the reaction plane are deflected to negative angles @xcite . at a certain intermediate incident energy , named the balance energy @xmath3 , the attractive component and the repulsive component of the interactions balance each other and consequently the flow crosses zero , changing from a negative sign at low energies to a positive sign at high energies .
the balance effect was extensively investigated at ganil @xcite and msu @xcite by measuring different colliding systems .
it was also studied in the framework of theoretical models @xcite . for semi - central collisions ,
the balance energy was found to be sensitive to the stiffness of the nuclear eos and to the in - medium reduction of the nucleon - nucleon cross section @xcite . at lower impact parameters
the balance energy is expected to be only sensitive to the in - medium nucleon - nucleon cross section @xcite .
recent experimental results @xcite revealed also , in the same beam energy range , signatures of a change in the azimuthal emission pattern of mid - rapidity particles , from an in - plane enhancement at low incident energies to the well known out - of - plane preferential emission , the so - called squeeze - out @xcite , at higher energies .
the incident energy where this transition takes place ( termed @xmath4 ) is also found to be sensitive to the in - medium nucleon - nucleon cross section @xcite .
the excitation functions of both sideward - flow and squeeze - out effects were measured with the fopi detector @xcite for the au+au system at seven incident energies between @xmath5@xmath0mev and @xmath6@xmath0mev .
the ability of the fopi device to detect intermediate mass fragments ( @xmath7 ) allows a cleaner identification of the collective flow signal and a better extrapolation of the measured excitation functions of the in - plane and of the out - of - plane flows toward their thresholds at low incident energies .
we found however that the threshold energies are rather insensitive to the charge of the considered particles ( @xmath8 ) .
we found also that the transition point seems to be located at higher beam energies than the balance point . on the other hand
, the fopi detector ensures a wide range of impact parameter collisions to be explored .
this offers the possibility to investigate the centrality dependence of @xmath3 and @xmath4 , an aspect which was recognized to be crucial in this kind of study @xcite .
this allows us to discuss for the first time this centrality dependence in conjunction with the threshold energy ( @xmath9 ) for the radial flow in highly central collisions .
it shows that the three threshold energies ( @xmath3 , @xmath4 , and @xmath9 ) might be attributed to a common phenomenon , the relevant parameter being the energy deposited into the nuclear system .
the data presented in this paper concern the au+au system at seven incident energies @xmath5 , 120 , 150 , 250 , 400 , 600 and 800@xmath0mev .
they have been collected with the phase i of the fopi detector @xcite at the sis / esr accelerator facility , gsi darmstadt . in its phase
i configuration , the fopi detector covers in full azimuth the laboratory polar angles ( @xmath10 ) from @xmath11 to @xmath12 .
it consists mainly of a highly segmented forward wall of plastic scintillators divided into two parts : the inner wall made of 252 trapezoidal scintillators which covers the @xmath10 range between @xmath11 and @xmath13 , and the outer wall made of 512 scintillator strips which covers the @xmath10 domain from @xmath14 to @xmath12 .
the forward wall provides an element identification and the velocity of the reaction products through energy loss and time - of - flight measurements .
its segment structure allows to determine the velocity vector components
. a complementary shell of 188 thin energy loss detectors is mounted in front of the forward wall in order to achieve lower detection thresholds .
this cluster detector is made of an ensemble of gas - filled ionisation chambers ( parabola ) mounted in front of the outer wall , and thin plastic scintillator paddles ( rosace ) combined with the inner wall . in order to reduce the background scattering in the air gas , a helium bag
is placed between the target and the detectors .
this setup measures simultaneously most of the light charged particles and intermediate mass fragments ( up to @xmath15 ) emitted in the forward center - of - mass ( c.m . ) hemisphere .
its high granularity allows high multiplicity events to be measured with a negligible multi - hit rate .
the apparatus ensures a very good azimuthal symmetry which is an important feature for the study of the flow phenomena .
the measured events were sorted according to their degree of centrality using the standard method based on the correlation between the multiplicity of emitted particles and the impact parameter .
the multiplicity distribution of the charged particles detected in the outer wall exhibits the typical plateau for intermediate values , followed by a steep decrease at the highest multiplicities @xcite .
the highest multiplicity bin ( named pm5 ) has been defined by cutting at half of the plateau value ( the corresponding lower limits of the pm5 multiplicity bin are given in tab .
[ tab1 ] ) .
the remaining part of the multiplicity distribution has been subdivided into four equally spaced bins ( named pm1 to pm4 ) according to the procedure used by the plastic ball collaboration @xcite .
the results presented in what follows include only events belonging to the pm3-pm5 multiplicity classes where background contamination , estimated from measurements without target , is negligible .
the mean impact parameter associated to each pm event class has been determined in the framework of the iqmd ( isospin quantum molecular dynamics ) model @xcite by filtering the theoretical calculations with the realistic simulator of the fopi detector . the resulting @xmath16 values and their r.m.s deviations are presented in tab .
[ tab2 ] for the beam energies @xmath17 and 400@xmath0mev .
note that @xmath16 values are the same within the r.m.s deviations at the other beam energies .
it can be seen from tab .
[ tab2 ] that the multiplicity criterium offers a large coverage of the impact parameter range . on the other hand , model studies @xcite ,
using the so - called quality factor introduced by cugnon and lhte @xcite , allowed us to show that with the fopi / phase - i setup , the multiplicity criterium , as compared to other criteria , appears as the most appropriate one for exploring flow observables over a large impact parameter domain .
the reaction plane was reconstructed with the transverse momentum analysis devised by danielewicz and odyniec @xcite . in order to remove autocorrelation effects ,
the azimuth of the reaction plane was estimated for each particle @xmath18 in a given event as the plane containing the vector @xmath19 and the beam axis , where @xmath19 is calculated from the transverse momenta @xmath20 of all detected particles except the particle i : @xmath21 @xmath22 is the multiplicity of the event and @xmath23 and @xmath24 otherwise .
@xmath25 is the @xmath26 particle rapidity divided by the projectile rapidity in the c.m .
the parameter @xmath27 , choosen equal to 0.5 , was introduced in order to remove mid - rapidity particles which have a negligible correlation with the reaction plane . according to @xcite , a boost velocity @xmath28 ( @xmath29 is the mass of the particle @xmath18 and @xmath30 is the sum of the projectile and target masses )
was applied to each particle @xmath31 in order to take into account the effects of momentum conservation due to the exclusion of the particle @xmath18 .
the influence of these effects on the observables considered in the present work was found to be of a few percent at the lowest beam energies ( @xmath5 and 120@xmath0mev ) and negligible at higher incident energies .
the accuracy of the reaction plane reconstruction ( _ i.e. _ the azimuthal deviation @xmath32 of the reconstructed reaction plane with respect to the true one ) was estimated for each event class by randomly dividing each event into two equal parts and by taking the one half of the angle between the @xmath33 vectors of the two subevents @xcite .
the azimuthal dispersion @xmath34 was found to vary typically from @xmath35 to @xmath36 for the pm event classes under consideration .
it is worth noting that the use of a heavy system such as au+au offers , in the beam energy range considered here , a good event characterisation both in centrality and reaction plane reconstruction as compared to lighter systems whose measured ejectile multiplicities are lower .
the in - plane flow component ( sideward - flow ) is examined in terms of the normalized in - plane transverse momentum @xmath37 ( @xmath38 where @xmath39 is the in - plane transverse momentum per nucleon and @xmath40 is the projectile momentum per nucleon in the c.m .
system ) as a function of the normalized c.m .
( @xmath41 is defined above ) .
this normalisation , suggested in earlier works @xcite , is motivated by the fact that one obtains a scale invariant representation of the data in a fluid dynamical description of the collision .
figure [ fig1 ] shows a typical example of @xmath37 ( fig .
[ fig1].a ) and @xmath42 ( fig .
[ fig1].b ) _ versus _
@xmath41 plots for @xmath43 particles detected in semi - central ( pm4 ) reactions at @xmath44@xmath0mev . since the experimental apparatus covers only the forward c.m .
hemisphere , the plot has been measured for positive @xmath41 rapidities and reflected for negative ones . as shown by fig .
[ fig1].b , the dependence of @xmath42 on @xmath41 exhibits the well known s - shape behavior @xcite demonstrating the collective transfer of momentum between the backward and the forward hemispheres .
the linear part of the curve in the participant region ( _ i.e. _ , at mid - rapidity ) reflects the so - called side - splash effect while the fall - off starting just below the projectile rapidity ( @xmath45 ) is caused by the bounce - off effect @xcite . a quantitative measure of the amount of flow in the participant region of the collision
is given by the so - called normalized flow parameter @xmath46 which is commonly defined as the slope of the @xmath42 _ versus _
@xmath41 curve at mid - rapidity @xcite : @xmath47 .
technically the @xmath46 parameter is obtained by fitting a polynomial function of the form : @xmath48 to the data ( fig .
[ fig1].b ) .
the fit was restricted to the linear branch of the s - shape curve . as shown in reference @xcite , because the particle momenta are not projected onto the true reaction plane , their projections are on average biased downward by a factor @xmath49 where @xmath32 , as mentioned before , is the estimate of the azimuthal deviation of the reconstructed reaction plane with respect to the true one .
the data shown in fig . [ fig1].b and all the @xmath46 values presented in what follows are corrected for this effect .
the correction factors ( @xmath49 ) were typically ranging from 1.10 to 1.45 depending on the beam energy and the multiplicity bin .
the precise evaluation of the acceptance effects on the in - plane flow is rather difficult due to the complexity of the different experimental constraints .
this can only be investigated in the framework of realistic simulations where theoretical calculations are passed through the detector filter . in this context
, we have used the iqmd model @xcite which is known to reproduce quite well experimental flow data @xcite .
a few thousand of iqmd events were generated over a large range of impact parameters with the hm choice ( hard eos plus a momentum dependent potential ) of the nuclear interaction .
this force is recognized as providing the best description of the observed trends in the in - plane flow data @xcite . for the present study ,
theoretical events were filtered applying geometrical cuts and energy thresholds of the fopi detector .
they were presorted in accordance with the above mentioned procedure used for the data ( see before ) .
the in - plane flow was extracted with respect to the true reaction plane which is known in the model .
because of the limited statistics , apparatus effects could only be evaluated for light particles .
we found that the experimental cuts only slightly affect @xmath42 values in the forward c.m .
the observed deviations are mainly caused by the geometrical limit of detection at @xmath50 .
this cut biases down the @xmath46 parameter by about @xmath51 for @xmath52 and by less than @xmath53 for heavier particles in the pm5 event class at @xmath44@xmath0mev .
note that these effects decrease with increasing fragment size @xcite because heavy particles , due to their low sensitivity to thermal fluctuations , occupy a smaller phase space .
therefore , in order to avoid misleading interpretations of the data , the @xmath46 flow parameters presented in what follows include only the measurements of particles whose charge is @xmath54 . on the other hand ,
it is worth noting that the effects of the @xmath50 cut decreases with increasing impact parameters since peripheral and semi - central event topologies are less accentuated in the transverse direction as compared to central events .
the out - of - plane flow component was investigated from the azimuthal distributions d@xmath55/d@xmath56
( @xmath56 is the azimuthal angle of the detected particle relative to the azimuth of the reaction plane ) around the beam axis of mid - rapidity particles , selected by imposing a rapidity cutoff @xmath57 .
it is now an established fact that the out - of - plane anisotropy increases strongly with the transverse momentum of charged particles @xcite .
therefore , in order to extract relevant information from the data , we determined the squeeze - out signal by choosing a @xmath58 cut which , within the acceptance , gives access to the largest momenta and sufficiently wide for statistics considerations .
this @xmath58 window is @xmath59 , @xmath60 being the particle transverse momentum per nucleon divided by the projectile momentum per nucleon in the c.m .
these rapidity and transverse momentum cuts used to extract the signal , define a portion of the phase space which is covered by the fopi detector acceptance @xcite .
figure [ fig2 ] shows a typical d@xmath55/d@xmath56 distribution for @xmath61 particles in the pm4 event class at an incident energy of @xmath44@xmath0mev .
a clear preferential emission is observed in the direction perpendicular to the reaction plane ( @xmath62 and @xmath63 ) .
this enhanced emission reflects the squeeze - out effect .
the magnitude of the latter is commonly defined as the ratio @xmath64 of the number of particles emitted perpendicular to the reaction plane to the number of particles emitted in the reaction plane @xcite : @xmath65 the @xmath64 ratio is extracted by fitting a function of the form @xmath66 to the data ( curve of fig .
[ fig2 ] ) .
thus @xmath64 is calculated as @xmath67 . according to this definition , @xmath68 and @xmath69
are related to a preferential emission of matter in the reaction plane and out of this plane , respectively while @xmath70 corresponds to a perfect azimuthally isotropic situation .
the anisotropy ratio @xmath64 can be corrected , like the flow parameter @xmath46 , for the uncertainties due to fluctuations of the reaction plane @xcite . however , we found with the help of simulations that , for the low multiplicity events , the values of @xmath71 , which are the quantities involved in these corrections , were not determined with good accuracy .
thus , @xmath64 ratios reported in what follows are not corrected for the effects of reaction plane fluctuations .
nevertheless , the possible influence of these effects on the observed trends will be discussed in the following .
figure [ fig3 ] shows the excitation functions of the scale invariant flow parameter @xmath46 for different particles ( @xmath72 2 to 5 ) .
a sudden decrease is observed in the incident energy region @xmath73@xmath0mev .
it is interesting to notice that this sudden change is much more pronounced in the case of the heavier fragments which are more sensitive to the collective motion .
an extrapolation with fermi functions allows us to estimate the balance energy ( intersection with the abscissa ) for different types of particles .
note that the balance energy values extracted from extrapolations of the data with other functions , such as logarithmic and second order polynomial ones , were found to be the same within error bars @xcite .
we have also verified that the use of other scaling variables for the in - plane transverse momentum , such as @xmath74 ( as used in @xcite ) or @xmath75 , leads to very similar results .
the extrapolation for @xmath61 fragments leads to an intersection energy of @xmath76@xmath0mev for events corresponding to the pm4 bin .
the resulting values for the other particles ( @xmath77 , 4 and 5 ) are the same within uncertainties as the one obtained for @xmath61 ( see tab . [ tab3 ] ) .
this confirms the observations established by studying lighter systems that the balance energy is independent of the size of the detected particle @xcite .
our present balance energy point is somewhat larger than the one obtained for the same system from other experiments @xcite .
this is probably due to the fact that our pm4 event class contains less central events than the one used in references @xcite . indeed , as it is shown in the following , the balance energy is found to decrease with decreasing impact parameters . with this in mind , the balance energy value extracted here is consistent with the systematics of the balance energy as a function of the mass of the combined projectile - target system obtained from msu and ganil results @xcite .
on the other hand , it is worth noting that for a heavy system like au+au , because of the strong coulomb repulsion , the overall force is always repulsive @xcite .
since the balance energy should correspond to the energy at which the attractive and repulsive component of the nuclear interaction balance each other , it must be evaluated without contamination of non nuclear contribution to the flow . for light systems
this problem is less severe because of the much weaker coulomb repulsion . nevertheless , for heavy systems one may hope to extract the correct value of the balance energy by extrapolating the @xmath46 values from sufficiently high energies where coulomb contribution is negligible .
the dependence of the anisotropy ratio @xmath64 on the collision impact parameter is presented in fig .
[ fig4 ] at four incident energies going from 100 to 400@xmath0mev .
the signal includes here the contributions of all detected particles each weighted by its charge . by doing
so we reconstruct a coalescence invariant quantity which makes meaningful the investigation of the anisotropy ratio as a function of the impact parameter and the beam energy .
the geometrical impact parameter @xmath78 was obtained from the measured multiplicity distributions by assuming a sharp - cut - off approximation .
this allows us to perform direct comparisons of data measured at different bombarding energies . as can be seen in fig .
[ fig4 ] , the correlation between the @xmath64 ratio and the impact parameter exhibits a very different trend as the incident energy decreases . at the highest bombarding energy ( 400@xmath0mev )
one can observe a bell - shaped distribution whose maximum is located at intermediate impact parameters ( close to @xmath79 ) .
it is worth noting that the results obtained at higher energies are quite similar to those observed at @xmath80@xmath0mev @xcite . with decreasing beam energies , the shape of the correlation evolves gradually toward a different trend which is an evidence for a clear change in the emission pattern .
thus at @xmath5@xmath0mev , with decreasing impact parameter , one observes a transition from a preferential in - plane emission ( @xmath81 ) to the squeeze - out effect characterized by an enhanced out - of - plane emission ( @xmath82 ) . before going to the interpretations of this behaviour
, it must be pointed out that two effects might influence the @xmath64 ratio : i ) the dispersion of the reconstructed reaction plane with respect to the true one , which tends to attenuate the magnitude of the signal and ii ) the sideward - flow deflection which favors the emission of particles in the reaction plane . in both cases
the magnitude of the effect is impact parameter and beam energy dependent .
therefore , in order to eliminate possible ambiguities in the interpretation of the experimental observations in fig .
[ fig4 ] , it was necessary to examine the respective influences of these effects on the correlation between the anisotropy ratio and the impact parameter .
a further complete analysis of the data has allowed us to show that the bell - like shape of the distribution observed at high beam energies is preserved after taking into account both mentioned effects @xcite . on the other hand ,
the transition from @xmath69 to @xmath68 at @xmath5@xmath0mev can not be caused by one of these two effects .
indeed , fluctuations of the reaction plane tend to attenuate an anisotropy signal regardless of whether the @xmath64 ratio is larger or smaller than 1 .
therefore , taking into account the corresponding corrections in fig .
[ fig4 ] , the transition effect would be even more pronounced . on the other hand , an extraction of the @xmath64 quantity around the flow axis would shift up the experimental points but at large @xmath78 s , where the in - plane enhancement is observed , the flow angle is expected to be fairly low @xcite in particular at the lowest beam energy ( @xmath5@xmath0mev ) which is close to the balance energy ( fig .
[ fig3 ] ) .
let us now go back to the interpretations of the experimental observations of fig .
[ fig4 ] . at high energies ,
the maximum located near @xmath83 is consistent with an expansion - shadowing picture , _
i.e. _ an expansion of the compressed matter in the central region of the collision which is hindered by the presence of cold spectator remnants .
it is worth noting that recent iqmd calculations for neutrons @xcite predict a similar bell - shaped correlation with a maximum around @xmath84 . at low incident energies
( @xmath85 150@xmath0mev ) , a clear evidence for a transition from an enhanced in - plane emission pattern to a preferential out - of - plane emission is observed when @xmath78 decreases .
the results show that this transition takes place close to @xmath86@xmath0mev for collisions with impact parameters @xmath87 .
this effect was already observed for a lighter system @xcite and very recently for the same system au+au @xcite .
it might be attributed to a change from a collective rotational behaviour governed by the attractive mean field at low energies , to the high energy squeeze - out effect resulting from the repulsive pressure built up during the high density stage of the collision @xcite .
the transition energy @xmath4 , corresponding to an azimuthally symmetric distribution ( @xmath88 ) , can be evaluated from the excitation function of the anisotropy ratio @xmath64 .
this has been already investigated in a previous analysis @xcite where the experimental @xmath4 values were compared to the predictions of the iqmd model .
the excitation functions of the @xmath64 ratio are presented in fig .
[ fig5 ] for @xmath77 and 3 particles emitted in semi - central ( pm4 ) collisions . since events were selected here over a large multiplicity bin , the large statistics allowed us to investigate the anisotropy signal in a reduced high @xmath58 window as compared to the previous one .
this @xmath58 condition ( fixed as @xmath89 ) was chosen in order to extract from the data the largest @xmath64 magnitudes within the acceptance of the detector .
as one can see from fig .
[ fig5 ] , the @xmath64 ratio tends to saturate above @xmath90@xmath0mev and seems possibly to decrease at higher energies . on the other hand
, the behaviour of the signal at low beam energies exhibits the same sudden change as in the case of the excitation functions of the sideward flow . in order to extract the transition energy , the data points in the beam energy range 100 - 400@xmath0mev
were fitted with a fermi function ( curves of fig .
[ fig5 ] ) .
the values of @xmath4 , reported in tab .
[ tab4 ] , were determined for @xmath77 and 3 particles at the intercept of @xmath70 with the curves .
note that the intersection energies were found to be quite insensitive to the form of the fitting function .
they seem to be , as observed in the case of the balance energy , independent of the type of the detected particle within the error bars .
it can be seen from the results presented in tab .
[ tab3 ] and tab .
[ tab4 ] that the transition energy for the pm4 event class ( @xmath91@xmath0mev ) is somewhat larger than the corresponding balance energy ( @xmath92@xmath0mev ) .
this finding agrees with recent theoretical calculations for the ca+ca system @xcite .
it could be explained by angular momentum effects or by the fact that at the balance point the compression is not high enough to generate an enhanced out - of - plane emission .
on the other hand , the @xmath64 ratio , which increases with the particle transverse momentum for @xmath93@xmath0mev @xcite , has been extracted over a narrow @xmath58 window ( @xmath94 ) .
this could suggest that the extracted transition energy is @xmath58 dependent . nevertheless , as shown by the insert in fig .
[ fig5 ] , at @xmath5@xmath0mev where the transition effect takes place , the @xmath64 ratio remains nearly close to 1 in the whole explored @xmath95 domain .
finally , as mentioned before , the @xmath64 ratios are not corrected for fluctuations of the reconstructed reaction plane because the @xmath71 values can not be accurately determined .
therefore , in order to estimate how the transition energy could be influenced by these fluctuations , we have extracted @xmath4 from the @xmath64 values obtained by correcting the measured anisotropy ratios using the @xmath71 factors calculated for filtered iqmd theoretical events , where the true reaction plane is known .
the resulting transition energy value , for @xmath61 particles , was found to be @xmath96@xmath0mev in the pm4 event class .
this value is very close to the one obtained without correction ( tab .
[ tab4 ] ) .
the balance and transition energies extracted from extrapolations of the measured @xmath46 and @xmath64 excitation functions in the three different pm multiplicity bins are shown in fig .
[ fig6 ] as a function of the collision centrality . @xmath3 and @xmath4 are expressed here not in terms of the projectile energy but of the corresponding energy in the c.m .
they were obtained , due to considerations of detector acceptance effects and statistics , from the measured excitation functions of @xmath61 for @xmath3 and @xmath77 for @xmath4 .
as it can be seen , the transition energy is larger than the balance energy over the whole explored impact parameter range . both threshold energies
@xmath3 and @xmath4 increase with increasing impact parameters .
such a behaviour has been also very recently observed in the case of the balance energy of the ar+sc system @xcite and is predicted by the iqmd model for the ca+ca system @xcite .
this increase of @xmath3 and @xmath4 with @xmath97 may indicate that the threshold of flow effects is related to the energy deposited locally into the overlap zone of the collision : if the local temperature is large enough , the generated pressure can overcome the attractive nuclear forces .
when going from peripheral to central collisions , the deposited energy becomes larger and the onset of flow is therefore expected to take place at a lower incident energy . at bombarding energies close to the fermi energy , the participant picture is not well developed like at higher energies and the local temperature depends sensitively on the local heat relaxation - time and on collision time . with this in mind
, it is interesting to compare the threshold energies in finite impact parameter collisions with the threshold energy for the radial expansion in central collisions .
the latter was determined in reference @xcite by extrapolating the excitation function of the mean radial flow velocity , measured in central collisions , toward low incident energies .
its value was found to be about @xmath98@xmath0mev @xcite which correponds to a c.m .
energy of @xmath99@xmath0mev .
note that this value is consistent with recent results for the au+au reaction obtained at @xmath100@xmath0mev @xcite .
now , considering again fig .
[ fig6 ] , a rough linear extrapolation of our balance and transition energy points toward @xmath16 = 0 leads to values of @xmath101@xmath0mev and @xmath102@xmath0mev , respectively .
these values are close to each other and consistent with the value of @xmath9 within the error bars .
it is still premature to go to more quantitative interpretations because of the large experimental uncertainties .
nevertheless , it is tempting to speculate on the basis of the observations in fig .
[ fig6 ] on a common phenomenon at the origin of radial flow , sideward - flow and squeeze - out effects which takes place for the same deposited energy : when nuclear matter is heated above a certain limit , the attractive nuclear forces are counterbalanced by the thermal pressure and the system starts to expand above this threshold at which the overall force becomes repulsive .
this expansion of nuclear matter tends to an azimuthally symmetric pattern ( radial flow ) in highly central collisions , while in semi - central collisions the expanding matter is pushed to the side ( sideward - flow ) and the presence of cold spectator remnants hinders participant nucleons to escape in the reaction plane , which gives rise to a preferential out - of - plane emission ( squeeze - out ) .
this scenario for the squeeze - out effect is consistent with the observed correlation between the anisotropy ratio and the impact parameter ( fig .
[ fig4 ] ) .
differences on the various threshold energies can be expected on the bases of effects like collision geometry , system size , coulomb contribution , angular momentum dissipation , non equilibrium dynamics , _
etc_. on the other hand , it is worth noting that the c.m .
energy of 8.7@xmath0mev where the radial flow sets in , is close to the one of the caloric curve , reported by the aladin collaboration @xcite , where the temperature grows strongly again after the presently debated plateau . as pointed out in ref .
@xcite , this observation suggests that the onset of flow could be interpreted as a possible signature of a liquid - gas phase transition .
indeed , as discussed before , the onset of flow phenomena indicates a change in the reaction scenario from a global repulsive mechanism to a global attractive mechanism when decreasing the incident energy . in a fluid dynamical vision of the collision , it could also be seen as a manifestation of a liquid - gas phase transition since this leads to a sudden decrease of the repulsive pressure and , consequently , of the flow magnitude .
collective effects of nuclear matter in au+au collisions at incident energies ranging from 100@xmath0mev to 800@xmath0mev were measured for light particles and intermediate mass fragments with the fopi detector at gsi . the centrality and beam energy dependence of both sideward - flow and squeeze - out effects were investigated .
the evolution of the squeeze - out magnitude @xmath64 with the impact parameter is found to change drastically with the incident energy . at high incident energies
the squeeze - out signal exhibits a bell - shape with a maximum located at intermediate impact parameters .
this trend is consistent with an expansion - shadowing scenario where the expansion of highly compressed participant nuclear matter is hindered by the presence of cold spectator remnants . at low energies ,
the data show clearly a transition from an in - plane preferential emission to an out - of - plane enhancement when the centrality increases . this phenomenon might be attributed to a change from a collective rotational motion for large impact parameters to the squeeze - out effect for smaller impact parameters @xcite .
the balance and transition energies , corresponding to the onset of sideward - flow and squeeze - out effects , respectively , have been evaluated from extrapolations toward lower beam energies of the excitation functions of the scale invariant flow parameter @xmath46 and the anisotropy ratio @xmath64 . both of them are found to be , within errors , independent of the size of the detected particle .
the transition energy is larger than the balance energy .
the extrapolation of the centrality dependence of both threshold energies toward @xmath103 leads to a value which is close to the threshold energy for the radial expansion in central collisions .
this suggests that the same phenomenon could be at the origin of the three processes .
the relevant parameter seems to be the energy deposited into the system in order to counterbalance the attractive nuclear forces although compression effects can also be present . in central collisions
the repulsive pressure is expected to be the highest and the participant matter can expand freely in all directions , while for @xmath104 , besides complex geometrical and dynamical effects , the presence of the spectator matter causes the appearence of sideward - flow and squeeze - out .
comparisons of the present experimental results with transport model predictions should provide interesting information about the in - medium reduction of the nucleon - nucleon cross section @xcite .
this work was supported in part by the french - german agreement between gsi and in2p3/cea and by the procope - program of the daad .
h. stcker , j.a .
maruhn and w. greiner , phys .
* 44 * ( 1980 ) 725 w. scheid , h. mller and w. greiner , phys .
* 32 * ( 1974 ) 741 see for example h.h .
gutbrod , a.m. poskanzer and h.g .
ritter , rep .
prog . phys .
* 52 * ( 1989 ) 1267 and references therein m.b .
tsang _ et al .
* 57 * ( 1987 ) 2720 j.p .
sullivan _ et al .
b * 149 * ( 1990 ) 8 j. peter , nucl .
phys . a * 545 * ( 1992 ) 173c w.q .
shen _ et al .
phys . a * 551 * ( 1993 ) 333 d. krofcheck _ et al .
rev . lett .
* 63 * ( 1989 ) 2028 c.a .
ogilvie _ et al .
c * 42 * ( 1990 ) r10 g.d .
westfall _ et al .
* 71 * ( 1993 ) 1986 r. pak _ et al .
c * 54 * ( 1996 ) 2457 j.j .
molitoris and h. stcker , phys . lett .
b * 162 * ( 1985 ) 47 h.m .
xu , phys .
* 67 * ( 1991 ) 2769 h.m .
xu , phys .
c * 46 * ( 1992 ) r389 v. de la mota _ et al . _ ,
c * 46 * ( 1992 ) 677 c. hartnack , thesis , frankfurt university , germany ( 1992 ) h.m .
xu , phys .
c * 47 * ( 1993 ) 891 d. klakov , g. welke and w. bauer , phys .
c * 48 * ( 1993 ) 1982 b.a .
li , phys .
c * 48 * ( 1993 ) 2415 h. zhou , h. li and y. zhuo , phys .
b * 318 * ( 1993 ) 19 y. ma , w. shen , j. feng and y. ma , z. phys . * 344 * ( 1993 ) 469 h. zhou , h. li and y. zhuo , phys .
c * 50 * ( 1995 ) r2664 s. soff _ et al .
c * 51 * ( 1995 ) 3320 e. lehmann _ et al .
_ , z. phys .
* 355 * ( 1996 ) 55 r. popescu _ et al . _ ,
b * 331 * ( 1994 ) 285 w.k .
wilson _ et al .
c * 51 * ( 1995 ) 3136 a. buta , m. petrovici , l. legrand , d. moisa , v. simion and the fopi collaboration , gsi report 95 - 1 ( 1995 ) p. 57 m.b .
tsang _ et al .
c * 53 * ( 1996 ) 1959 h.h .
gutbrod , k.h .
kampert , b.w .
kolb , a.m. poskanzer , h.g .
ritter , and h.r .
schmidt , phys .
b * 216 * ( 1989 ) 267 m. demoulins _ et al .
b * 241 * ( 1990 ) 476 a. gobbi and the fopi collaboration , nucl .
inst . meth .
a * 324 * ( 1993 ) 156 j.p .
alard and the fopi collaboration , phys .
* 69 * ( 1992 ) 889 k.g.r .
doss _ et al .
c * 32 * ( 1985 ) 116 c. hartnack , j. aichelin , h. stcker and w. greiner , phys .
b * 336 * ( 1994 ) 131 p. crochet , thesis , strasbourg university , france ( 1996 ) j. cugnon and d. lhte , nucl .
phys . a * 397 * ( 1983 ) 519 p. danielewicz and g. odyniec ,
b * 157 * ( 1985 ) 146 c.a .
ogilvie _ et al .
c * 40 * ( 1989 ) 2592 n. balzs , b. schrmann , k. dietrich and l.p .
csernai , nucl .
phys . a * 424 * ( 1984 ) 605 a. bonasera and l.p .
csernai , phys .
* 59 * ( 1987 ) 630 h.a .
gustafsson _ et al .
* 52 * ( 1984 ) 1590 k.g.r .
doss _ et al .
* 57 * ( 1986 ) 302 f. rami _ et al .
_ , xxxi international winter meeting on nuclear physics , bormio ( 1993 ) p. 200 v. ramillien and the fopi collaboration , nucl .
a * 587 * ( 1995 ) 802 t. wienold , thesis , heidelberg university , germany ( 1993 ) p. crochet and the fopi collaboration , in preparation r. don and the fopi collaboration , in preparation w. reisdorf and the fopi collaboration , nucl .
phys . a * 612 * ( 1997 ) 493 p. crochet and the fopi collaboration , gsi report 95 - 1 ( 1994 ) p. 51
d. brill _ et al .
_ , z. phys .
a * 355 * ( 1996 ) 61 n. bastid and the fopi collaboration , accepted for publication in nucl .
a l. lambrecht _ et al .
_ , z. phys .
a * 350 * ( 1994 ) 115 h.h .
gutbrod , k.h .
kampert , b.w .
kolb , a.m. poskanzer , h.g .
ritter , r. schicker and h.r .
schmidt , phys .
c * 42 * ( 1990 ) 640 w.m .
et al . _ ,
c * 42 * ( 1990 ) r491 m.d .
partlan _ et al .
* 75 * ( 1995 ) 2100 a. buta _ et al .
phys . a * 584 * ( 1995 ) 397 p.
_ , xxxiv international winter meeting on nuclear physics , bormio ( 1996 ) p. 421 s.a
. bass , c. hartnack , h. stcker and w. greiner , preprint , gsi-94 - 12 ( 1994 ) a. gobbi , xiv international joliot - curie school of nuclear physics , maubuisson , france ( 1995 ) m. dagostino _ et al .
b * 368 * ( 1996 ) 259 j. pochodzalla _ et al .
* 75 * ( 1995 ) 1040 g. papp and w. nrenberg , ahp heavy ion physics * 1 * ( 1995 ) 241 | using the fopi detector at gsi darmstadt , excitation functions of collective flow components were measured for the au+au system , in the reaction plane and out of this plane , at seven incident energies ranging from 100@xmath0mev to 800@xmath0mev .
the threshold energies , corresponding to the onset of sideward - flow ( balance energy ) and squeeze - out effect ( transition energy ) , are extracted from extrapolations of these excitation functions toward lower beam energies for charged products with @xmath1 .
the transition energy is found to be larger than the balance energy .
the impact parameter dependence of both balance and transition energies , when extrapolated to central collisions , suggests comparable although slightly higher values than the threshold energy for the radial flow .
the relevant parameter seems to be the energy deposited into the system in order to overcome the attractive nuclear forces .
+ * keywords :* heavy ion collisions , nuclear matter expansion , sideward - flow , squeeze - out , radial flow , balance energy , transition energy . |
a few tens of bright , young , and massive stars have been observed in close proximity ( @xmath3@xmath4 pc ) to the radio source sgra * , which is associated with the supermassive ( @xmath5 ) black hole in the galactic center ( see genzel et .
al . 2000 , ghez et .
al . , 2003 ,
shodel et .
2002,2003 for recent work ) .
the apparent youth of these stars implies two logical possibilities for their birth history : ( 1 ) either they were born _ in situ _ , in the immediate vicinity of sgra * , or ( 2 ) they were born some distance away and then delivered to sgra * within a few million years after their birth . in the first scenario , the _ in situ
_ birth would occur in a strong tidal field of sgra * , and the density of the star - forming gas would have to be several orders of magnitude greater than that anywhere else in the galaxy ( phinney 1989 , sanders 1992 , morris 1993 ) . the most likely geometry for the star - forming gas is a dense disc , which is either compressed by the central explosion powered by sgra * ( morris , ghez , and becklin 1999 ) , or , more plausibly , which is massive enough to become self - gravitating and fragment into stars ( levin and beloborodov 2003 , nayakshin , cuadra , and sunyaev 2004 , milosavljevic and loeb 2004 , nayakshin and cuadra 2004 ) .
the _ in situ _ formation is not the focus of this paper , although we will briefly return to it in our concluding remarks .
the second scenario , which is the focus of this paper , calls for rapid delivery of the young stars to the supermassive black hole .
the initial version of it was proposed by gerhard 2001 , and investigated in more detail by mcmillan and portegies zwart 2003 , and kim and morris 2002 .
gerhard s basic idea was that if a massive ( @xmath6 ) cluster of stars was born @xmath7@xmath8pc away from sgra * , then the dynamical friction in the central bulge would bring the cluster towards sgra * on the timescale of a few million years . during the inward migration
, the massive stars in the cluster would sink to the cluster s center and form a compact core .
gerhard has argued that even though the envelope of the cluster gets stripped by the tidal field a few parsecs away from sgra * , the dense cluster core would only be tidally disrupted within the central parsec .
however , even that is not close enough : the observed stars reside @xmath3@xmath4 pc from sgra * and the core density would have to increase by orders of magnitude before it could get so close to sgra*. recently , hansen and milosavljevic 2003 ( from now on , hm ) have suggested a fix to gerhard s scenario which would enable the stars to be delivered to the observed proximity from sgra*. they have argued that if an intermediate - mass black hole ( imbh ) , with the mass of @xmath9@xmath10 , was positioned at the center of the cluster , than it would provide some extra gravitational binding for the cluster core .
this scenario is supported by an observation of the seemingly bound mini - cluster irs13 at @xmath11pc from sgra * ( maillard et .
2004 , from here on mpsr ) .
mpsr have argued that the presence of an imbh at the center of irs13 is the most plausible explanation of how the mini - cluster keeps itself together in the strong tidal field of sgra*. on the theoretical side , a number of investigations argue that imbh would plausibly form in a dense core of a young very massive cluster ( portegies zwart et .
2004 , gurkan et .
al . 2004 ) .
the cluster+imbh inspiral would consist of two stages : the first one , in which most of the most of the cluster is tidally disrupted at about a parsec from sgra * and the imbh is released together with a few tens of massive stars tightly bound to it , and the second one , in which the imbh continues its inspiral while the stars are peeled off one - by - one by the sgra * tidal pull , and are eventually put into nearly keplerial orbits around sgra*. the first stage was recently investigated numerically by kim and morris 2003 ( km ) and by gurkan and rasio 2004 .
km have found that very few stars survive bound to imbh once it gets to the central parsec . however , one of the numerical limitations in km ( pointed out by the authors ) was the large softening length for gravitational interactions between the stars .
thus km did not model faithfully the internal dynamics of the cluster .
a more exact monte - carlo - based simulation of gurkan and rasio 2004 has found that the number of stars bound to imbh at the end of the first stage is consistent with the number of young stars in the central half - parsec .
here we focus on the second stage the final inspiral of the imbh .
we investigate numerically the stellar orbits produced by such an inspiral , and compare them to observations . in the next section
we describe the algorithm which was used for the numerical orbital integrations ; this section may be skipped by a reader uninterested in technical details . in section 3
we present our numerical results , and in section 4 compare them to the data .
we find that while the inspiral on a circular orbit explains nicely the clockwise - moving stellar disc in the galactic center ( levin and beloborodov 2003 , genzel et .
al . 2003 ) , it does not reproduce the thick counterclockwise stellar disc and the randomly oriented tight orbits of the s - stars ( ghez et .
2003 , shodel et .
al . 2003 ) .
the inspiral on an eccentric orbit does push some orbits out of the inspiral plane , but is still not entirely consistent with the current data .
we also show that the proposed imbh at the center of irs13 almost certainly could not have delivered all of the young stars to the galactic center .
finally , we speculate how the hm scenario may be modified to account more closely for all of the kinematic data .
the stars in our system will experience repeated close encounters with the imbh even after they get tidally stripped from it , and we expect the resulting orbits to be very chaotic and sensitive to the initial conditions
. therefore , our study will be statistical in nature : we are seeking to find the _ distribution _ of orbits which are produced by the imbh inspiral .
symplectic algorithms ( sa ; see yoshida 1990 ) are particularly suitable for statistical studies , since they generate trajectories which conserve the phase - space volume exactly .
integrators which use sas have excellent long - term behavior , since the hamiltonian mapping used in a sa is integrated exactly and is only slightly different from a true hamiltonian mapping of the system .
however , for our purposes the simplest symplectic codes have a disadvantage since they do not handle well the close encounters between a star and imbh ( see preto and tremaine 1999 for discussion ) .
the origin of this limitation is the reduced timestep which is used to integrate through the encounter ; reduction of the timestep leads to change of the sa hamiltonian and thus may cause spurious energy changes . to overcome this problem , mikkola ( 1997 ) , preto and tremaine ( 1999 ) , mikkola and tanikawa ( 1999 ) , and mikkola and wiegart ( 2002 ) have developed _ extended phase - space _ symplectic integrators which use fixed timestep but in an extended phase space .
the basic idea for extending the phase space of the hamiltonian system goes back to poincare , and can be briefly described as follows : let @xmath12 be a time - dependent hamiltonian , where @xmath13 and @xmath14 stand for spatial coordinates and their conjugate momenta respectively and @xmath15 is the time .
let us extend the phase space of the system by another coordinate @xmath16 and its conjugate momenta @xmath17 .
consider now a new hamiltonian @xmath18 in the extended phase space ( mikkola 1997 ) : @xmath19 , \label{gamma1}\ ] ] where @xmath20 is some smooth differentiable function .
this hamiltonian depends only on the coordinates and momenta of the extended phasespace ; thus any trajectory in the extended phase space which satisfies hamiltonian equations will conserve @xmath18 .
therefore , if a particle begins its motion on the hypersurface @xmath21 , it will always stay on this hypersurface .
moreover , the trajectory of this particle @xmath22 $ ] will trace the values @xmath23 $ ] corresponding to the trajectory of the original hamiltonian system ; here @xmath24 is the time variable which marks evolution in the extended phase space .
the two time variables are related by @xmath25 the idea then is to integrate the system numerically in the extended phase space using the constant timestep @xmath26 , but choosing @xmath20 so that the real timestep @xmath27 is small during a close encounter . to integrate the motion in a symplectic way
, we need the hamiltonian @xmath18 to be separable into two parts , @xmath28 so that the motion due to each of @xmath29 and @xmath30 would be integrable analytically . once such separation is found , the system can then be integrated via a generalized leapfrog step : @xmath31 where @xmath32 and @xmath33 represent the forward motion in the extended phasespace by the time interval @xmath26 under the action of @xmath29 and @xmath30 , respectively .
preto and tremaine 1999 and mikkola and tanikawa 1999 have independently found the form of the generalized hamiltonian which is the sum of two analytically integrable parts : @xmath34 ) , \label{ptr}\ ] ] where @xmath35 and @xmath36 are the kinetic and potential energies , respectively , and @xmath37 is an arbitrary smooth function .
the hamiltonian in eq .
( [ ptr ] ) is obviously of the form in eq .
( [ gamma1 ] ) , with @xmath38 .
when @xmath39 , i.e. at the hypersurface representing the original hamiltonian system , @xmath40 .
the choice @xmath41 possesses the following surprising and useful property : the generalized leapfrog step for a two - body problem follows _ exactly _ the correct conic - section orbit , albeit with some time error . the real time and the time in the extended phase space
are then related by @xmath42 we can now see the advantage of using the extended phase space . in the extended phase space , we can integrate the motion by using the generalized leapfrog in eq .
( [ u ] ) with the timestep fixed , which is required for energy conservation .
however , the timestep in real phase space will not be constant : when the potential energy becomes large during a close encounter , the timestep in real space becomes small and the precision of real - space integration increases . for our problem however , the generalized hamiltonian in eq .
( [ ptr ] ) is not optimal .
we are interested in a restricted 3-body problem where a star is interacting with two much heavier black holes .
the mass ratio between sgra * and the imbh is @xmath43 or more% of the cluster s mass . ] and our star is initially bound to imbh .
the potential energy of the star per its unit mass is given by @xmath44 where @xmath45 is the position vector of the star , @xmath46 and @xmath47 are the mass and the position vector of sgra * , and @xmath48 and @xmath49 are those of the imbh . in our simulations ,
the star is treated as a massless particle and the orbits of the two black holes are introduced by hand , with the inspiral prescription motivated by analytical calculations .
we see from eq .
( [ u2 ] ) that the potential energy of the star is dominated by its interaction with sgra * , even when it is well inside the roche lobe of the imbh . thus if the hamiltonian of eq .
( [ ptr ] ) is used to generate the integrator , the following paradoxical situation may occur : the star may be on a bound orbit around the imbh or it may be experiencing a close encounter with the imbh , yet the choice of timestep will be dominated by the star s interaction with sgra * and not with the imbh .
clearly , to make the integrator efficient , we need to reduce the relative contribution to the timestep from the star
sgra * interaction .
the following generalized hamiltonian , found by mikkola and wiegert in 2002 , does the trick : @xmath50 here all the position vectors are measured relative to the barycenter of black - hole binary , which is very close to sgra * , @xmath51 is a positive number , and the term @xmath52 is given by @xmath53 since we treat the star as a test particle and introduce the orbits of both black holes by hand , the vectors @xmath47 and @xmath49 are time - dependent and therefore are prescribed functions of @xmath16 .
both components of the hamiltonian in eq .
( [ mw ] ) are easily integrable : the first term on the rhs describes keplerian motion around the barycenter and the second term describes the motion where all positions stay constant and all momenta change linearly with time ; therefore we can efficiently construct the generalized leapfrog operator for this hamiltonian .
the timestep in real phase space is then given by @xmath54 if one chooses @xmath55 , then at the imbh roche surface the star - sgra * interaction and star - imbh interaction contribute approximately equally to the timestep , and the contribution from the @xmath52-term is relatively small .
mikkola and wiegert ( 2002 ) have proposed to use the generalized hamiltonian in eq .
( [ mw ] ) for simulating the motion of near - earth and near - jupiter asteroids .
they have cautioned however , that if the asteroid gets too close to the sun , the @xmath52-term in eq .
( [ mw ] ) may become large , and the hamiltonian may become singular .
our simulations have confirmed this .
we have found that while for circular inspiral the hamiltonian in eq .
( [ mw ] ) always works well , for an eccentric inspiral some of the simulation runs crash because the hamiltonian becomes singular .
however , we have identified a way to fix this singularity problem by a slight modification of both parts of the generalized hamiltonian : @xmath56 the above hamiltonian is of the form in eq .
( [ gamma1 ] ) and , moreover , the two terms on the rhs are still separately integrable .
the latter statement is not trivial for the first term on the rhs ; however , recall that for a particle motion in a spherically symmetric potential @xmath57 the effective potential for the radial motion is given by @xmath58 , where @xmath59 is the particle s angular momentum . thus adding the term @xmath60 to the hamiltonian is equivalent to rescaling of the particle s angular momentum in the radial equation of motion : @xmath61 , and solving the equations of motion with this term can be reduced to solving a purely keplerian problem .
it is straightforward to show that by setting the value of @xmath62 so that @xmath63 one can avoid the singularity in the generalized hamiltonian ; here @xmath64 is the maximal distance from the imbh to the barycenter during the simulation run . in this work ,
we have used the hamiltonians in eq .
( [ mw ] ) and eq .
( [ yl ] ) to simulate the circular and eccentric imbh inspirals , respectively .
we have also , as a check , have run test simulations with symba , a version of widely available swift code with the added ability to resolve close encounters between massive objects ( duncan , levison , and lee 1998 ) .
we have confirmed that our integrator and symba give results which are consistent with each other .
the efficiency of our code has allowed us to run thousands of inspiral simulations and obtain a distribution of the orbital parameters of the stars at the end of the inspiral . in the next section ,
we discuss the results of our simulations .
we show in the appendix that the eccentricity of the inspiraling imbh remains nearly constant for the observed density profile of the stellar cluster .
the semimajor axis evolves with time as @xmath65 where @xmath66 is the characteristic inspiral time given by @xmath67 where @xmath46 is the mass of sgra * , @xmath48 is the mass of the imbh , @xmath68 is the mass density of stars in the central cluster , and @xmath69 is the coulomb logarithm . for circular orbit
, this can be rewritten as @xmath70 where @xmath71 is the orbital period of the imbh .
genzel et .
2003 ) have measured the density profile of the central cusp to be @xmath72 for a realistic imbh @xmath73 .
therefore for the stars located a few arcseconds from sgra * the relevant inspiral timescale is between a few hundred and a few thousand orbital periods . in our simulations we use fiducial values of the mass ratio @xmath74 and the inspiral timescale of one thousand initial orbital periods of the imbh .
our inspiral timescale remains constant over the simulation run ; this approximation would be exact if the cusp density profile scaled as @xmath75 ( see e.g. appendix of gould and quillen , 2003 ) .
we spot - check our results by varying the inspiral timescale and making sure that our qualitative conclusions do not change .
figure 1 shows a pictorial representation of a typical circular inspiral .
the points @xmath76 on the figure represent ten thousand snapshots of the @xmath77 and @xmath78 coordinate of the star .
the orthogonal rotating axis @xmath77 and @xmath78 belong to the inspiral plane and are chosen so that the sgraa * and imbh are always on the @xmath77 axis . from figure 1
we can trace the dynamical history of the star as the inspiral proceeds .
initially the star is bound to the imbh which is climbing towards the center on the @xmath77-axis .
then the tidal field of sgra * disrupts the star - imbh binary , the star escapes from the imbh via the inner lagrange point and spends some time inside of the imbh orbit .
then as a result of a close encounter , the star is flung outside of the imbh orbit .
the star then continues to have encounters with the imbh , and its eccentricity and inclination get kicks during the encounters .
eventually , the orbit of the imbh is shrunk enough so that the encounters do not occur any more and the star s eccentricity and inclination stay virtually constant . figure 2 and 3 show how the star s eccentricity and inclination evolve with time . using cita s mackenzie cluster ,
we have performed one thousand simulation runs , such as the one shown in figures 1,2 , and 3 .
the initial conditions were different for each run ; we have chosen the stars in all the runs to have the same initial jacoby integral . in figure 4
we show in the @xmath79 plane the scatter plot of the final eccentricities and inclinations . because of multiple close encounters , the orbits are highly chaotic : even a slight change in the initial conditions or in the timestep of the integrator leads , after some time , to a very different trajectory and thus to a different final eccentricity and inclination relative to the inspiral plane .
one subtle issue is the choice of timestep in the extended phase space ; see the previous section . ] .
usually , one can make sure that the chosen timestep is appropriate by running the simulation with the fraction of the chosen timestep and checking if the simulation results stays the same . in our case
however , the system is chaotic and the individual trajectory will always be different after some time if a different integration timestep is chosen .
however , it is not the individual trajectories that we are after ; rather , we want to know what _ distribution _ of the orbital parameters one should expect in the inspiral scenario
. therefore , we have repeated the @xmath43 runs with the same initial conditions but with the timestep of @xmath80 of the original timestep , and checked if the two sets of @xmath79 points could belong to the same two - dimensional distribution . the first thing to check was whether the averages agreed ; they did .
however , we wanted a more complete test .
for one - dimensional data , kolmogorov - smirnov test is both popular and mathematically well justified . for two - dimensional data ,
there exist only semi - empirical algorithms based on monte - carlo simulations : see peacock ( 1983 ) and fasano and franceschini ( 1987 ) .
we used the procedure outlined in the latter to check whether the two @xmath79 data sets were compatible with the same distribution ; they were .
the remarkable thing about the scatter plot in fig . 4
is that while the stars , on average , end up with significant eccentricity , more than 99% of them have inclinations smaller than 10 degrees .
we make another 1000 runs with the inspiral rate reduced by a factor of @xmath7 , we find that while the average eccentricity has increased significantly , and the distribution of inclinations remains similar , with only a few greater than 10 degrees .
the final inclinations are no longer small when the inspiraling imbh is on an eccentric orbit . qualitatively , eccentric stellar orbits experience secular torque if the semimajor axes of the star and imbh are misaligned . hence angular momentum , inclination , and eccentricity of the star can experience substantial secular changes during the course of the inspiral . in figure 5
we plot a fraction of stars with inclinations greater than 10 degrees for a few values of imbh eccentricity .
each point of the graph is obtained by simulating 100 stellar orbits , and the stars which become unbound from sgra * are removed from the data set . in figure 6 , we show average inclination of stellar orbits after the inspiral , for the same values of imbh eccentricity . in figure 7
, we show the scatter plot of final inclinations and eccentricities of stars for the case when the imbh eccentricity is @xmath81 ; the scatter plot contains @xmath43 points .
we observe two groups of unejected stars which are roughly equal in size . in one group ,
the inclinations are less than 10 degrees , and in another the inclinations are spread evenly between 10 and 180 degrees . in figures 8 and 9
we show the time evolution of inclination and eccentricity for a typical star which ends up with high inclination .
current kinematic data indicates that there are two nearly orthogonal stellar discs in the galactic center .
one of the disks consists of clockwise - moving stars at @xmath82 from the black hole , and has dispersion in inclinations no larger than 10 degrees ( levin and beloborodov 2003 , genzel et .
al . 2003 ) .
some of the stars in this clockwise disc have significant eccentricities ( beloborodov and levin , 2005 ) .
the second disc consists of stars which move counter - clockwise ( genzel et .
al . 2003 ) and its thickness has not been estimated yet ; however it is probably thicker than the clockwise disc since the planar fit to its stellar velocity vectors has a large @xmath83 of @xmath84 .
the mini - cluster irs13 with a putative imbh belongs to this second , thicker disc . in our simulations
the circular inspiral nicely reproduces the kinematics of the thin clockwise disc : dispersion of inclinations is less than 10 degrees yet some of the stars possess significant eccentricities .
however , the nearly circular inspiral would never produce the second population of stars which move counter - clockwise .
eccentric inspiral seems to be more promising : while significant fraction of stars remain close to the inspiral plain , the rest of the stars get pushed out of the inspiral plain and end up with large inclinations ; see figures 5 and 7 .
however , the stars with large inclinations do not have a preferred orbital orientation , thus it is hard to imagine them assembling a second disc . upcoming observations will better constrain the thickness of the second disc and then it will become possible to make a more quantitative assessment of how likely it is to produce the observed counter - clockwise trajectories by an imbh inspiraling through the plane of the clockwise disc .
it seems certain , though , that an imbh in a counter - clockwise moving irs13 would not be able to produce the thin clockwise disc .
thus the presence of imbh in irs13 could not account for kinematics for all of the young stars at @xmath11pc from sgra*. the imbh in irs13 would also have difficulty generating highly compact eccentric orbits of the most central stars like so-2 or so-16 ( see ghez et .
al . 2005 ) .
these stars apoapses are an order of magnitude closer to sgra * than irs13 .
such disparity of stellar and imbh orbits is never observed in our simulations .
however , we note from figure 9 that in the case of an eccentric inspiral , the test bodies with high inclination undergo secular evolution in which their eccentricity reaches values close to 1 , and their distance of closest approach to the central hole is much smaller than their semimajor axis . thus ,
if such a test body were a binary , this binary could get disrupted by sgra * , and one of the components of the binary could remain tightly bound to sgra*. the binary - disruption scenario for formation of the so - stars was already proposed by gould and quillen ( 2003 ) in a different context .
we thank scott tremaine for introducing us to symplectic integrators in the extended phase space , and carlo contaldi for showing us how to use the cita mckenzie cluster .
we have benefited from discussions with andrei beloborodov , jeremy goodman , brad hansen , chris matzner , milos milosavljevic , and especially with norm murray .
our research was supported by nserc .
the orbit of an imbh in the galactic center shrinks due to dynamical friction , and its eccentricity may also evolve with time .
the details of this evolution depend on the structure of the stellar cusp surrounding the black hole .
it is instructive and convenient to model the cluster by a collection of stars of mass @xmath85 with an isotropic distribution function @xmath86 where @xmath87 is the binding energy of the star , and @xmath88 and @xmath89 are real numbers . for a keplerian orbit ,
the angular momentum and binding energy are given by @xmath90 and @xmath91 , respectively ; here @xmath92 is the eccentricity of the orbit . from these relations
it is straightforward to derive the evolution equation for imbh eccentricity : @xmath93 here the symbol @xmath94 represents averaging over the orbital period .
the acceleration from dynamical friction acting on the imbh with velocity vector @xmath95 is given by the chandrasekhar formula @xmath96 where @xmath97 therefore , @xmath98 and @xmath99 cf .
eqs ( a5 ) and ( a6 ) of gould and quillen ( 2003 ) .
it is convenient to perform orbital averaging by expressing all the quantities through the mean anomaly @xmath100 : @xmath101 after some algebra , one gets @xmath102 where @xmath103^{\gamma}(1-e\cos\phi)^{1-\gamma } , \label{k1}\ ] ] and @xmath104^{\gamma}(1-e\cos\phi)^{-\gamma } ( 1+e\cos\phi ) .
\label{k2}\ ] ] we evaluate the integrals numerically and in fig .
10 we plot @xmath105 as a function of eccentricity , for @xmath106 and @xmath107 .
the two values of @xmath89 correspond to the observed density profile @xmath108 and the bahcall - wolf density profile @xmath109 .
we see from the graphs that in both cases eccentricity changes by a factor less than 2 as the semimajor axis shrinks by 3 orders of magnitude .
hence , in our simulations we are justified in keeping the eccentricity of imbh orbit constant .
alexander , t. , & livio , m. 2004 , 606 , l21 duncan , m. j. , levison , h. f. , & lee , m. h. 1998 , aj , 116 , 2067 fasano , g. , & franceschini , a. 1987 , mnras , 225 , 155 genzel , r. , et .
al . 2000 ,
mnras , 317 , 348 genzel , r. , et .
2003 , apj , 594 , 812 gerhard , o. 2001 , apj , 546 , l39 ghez , a. m. , et .
2003 , apj , 586 , l127 ghez , a. m. , et .
2005 , to appear in apj ( astro - ph/0306130 ) goodman , j. , & tan , j. c. 2004 , apj , 608 , 108 gould , a. , & quillen , a. c. 2003 , apj , 592 , 935 gurkan , m. a. , freitag , m. , & rasio , f. a. 2004 , apj , 604 , 632 gurkan , m. a. , & rasio , f. a. 2004 , submitted to apj ( astro - ph/0412452 ) hansen , b. m. s. , & milosavljevic , m. 2003 , apj , 593 , l77 jaroszynski , m. 2000 , aca , 50 , 67 kim , s. s. , figer , d. f. , & morris , m. 2004 , apj , 607 , l123 kolykhalov , p. i. , & sunyaev , r. a. 1980 , sval , 6 , 357 levin , y. , & beloborodov , a. m. 2003 , apj , 590 , l33 levin , y. 2003 , astro - ph/0307084 maillard , j. p. , et . al .
2004 , a&a , 423 , 155 mcmillan , s. l. w. , & portegies zwart , s. f. 2003 , apj , 596 , 314 mikkola , s. 1997 , cemda , 67 , 145 mikkola , s. , & tanikawa , k. 1999 , cemda , 74 , 287 mikkola , s. , & wiegert , p. 2002 , cemda , 82 , 375 milosavljevic , m. , & loeb , a. 2004 , apj , 604 , l45 morris , m , ghez , a. m. , & becklin , e. e. 1999 , advances in space research , 23 , 959 nayakshin , s. , cuadra , j. , & sunyaev , r. 2004 , a&a , 413 , 173 peacock , j. a. 1983 , mnras , 202 , 615 phinney , e. s. 1989 , the center of the galaxy ( 136th symposium of the iau , edited by m. morris ) , 543 portegies zwart , s. f. , et .
2004 , nature , 428 , 724 preto , m. , & tremaine , s. 1999 , aj , 118 , 2532 sanders , r. h. 1998 , mnras , 294 , 35 schodel , r. , et .
al . 2002 ,
nature , 419 , 694 schodel , r. , et .
2003 , apj , 596 , 1015 shlosman , i. , & begelman , m. 1987 , nature , 329 , 810 yu , q. , & tremaine , s. 2003 , apj , 599 , 1129 | many young stars reside within the central half - parsec from sgra * , the supermassive black hole in the galactic center . the origin of these stars remains a puzzle .
recently , hansen and milosavljevic ( 2003 , hm ) have argued that an intermediate - mass black hole ( imbh ) could have delivered the young stars to the immediate vicinity of sgra*. here we focus on the final stages of the hm scenario .
namely , we integrate numerically the orbits of stars which are initially bound to the imbh , but are stripped from it by the tidal field of sgra*. our numerical algorithm is a symplectic integrator designed specifically for the problem at hand ; however , we have checked our results with symba , a version of the widely available swift code .
we find that the distribution of the post - inspiral orbital parameters is sensitive to the eccentricity of the inspiraling imbh .
if the imbh is on a circular orbit , then the inclinations of numerically computed orbits relative to the inspiral plane are almost always smaller than 10 degrees , and therefore ( a ) the simulations are in good agreement with the observed motions of stars in a clockwise - moving stellar disc , ( b ) the simulations never reproduce the orbits of stars outside this disc , which include those in the second thick ring of stars and the randomly oriented unrelaxed orbits of some of the s - stars . if the imbh s orbital eccentricity is @xmath0 , then approximately half of the stars end up with orbital inclinations below 10 degrees , and another half have inclinations anywhere between @xmath1 and @xmath2 degrees ; this is somewhat closer to what s observed .
we also show that if irs13 cluster is bound by an imbh , as has been argued by maillard et .
al .
2004 , then the same imbh could not have delivered all of the young stars to their present location . |
small covers are a class of particularly nicely behaving manifolds @xmath4 , introduced by davis and januszkiewicz @xcite , each of which is an @xmath5-dimensional closed manifold with a locally standard @xmath6-action such that its orbit space is a simple convex @xmath5-polytope @xmath7 .
there are strong links of small covers with combinatorics and polytopes .
davis and januszkiewicz showed that small covers have very beautifully algebraic topology .
for example , the equivariant cohomology of a small cover @xmath8 is exactly isomorphic to the stanley - reisner face ring of @xmath7 , and the mod 2 betti numbers @xmath9 of @xmath10 agree with the @xmath11-vector @xmath12 of @xmath7 .
in addition , they also showed that each small cover @xmath8 determines a characteristic function @xmath13 ( here we call it a @xmath6-coloring ) on @xmath7 , defined by mapping all facets ( i.e. , @xmath14-dimensional faces ) of @xmath7 to nonzero elements of @xmath6 such that @xmath5 facets meeting at each vertex are mapped to @xmath5 linearly independent elements , and conversely , up to equivariant homeomorphism , @xmath10 can be reconstructed from the pair @xmath15 .
more specifically , take a point @xmath16 in the boundary @xmath17 , then there must be a @xmath18-dimensional face @xmath19 of @xmath7 such that @xmath16 is in the relative interior of @xmath19 , where @xmath20 . since @xmath7 is simple ( i.e. , the number of facets meeting at each vertex is exactly @xmath5 ) , there are @xmath21 facets @xmath22 such that @xmath23 .
let @xmath24 denote the rank-@xmath25 subgroup of @xmath26 determined by @xmath27 .
then we can define an equivalence relation @xmath28 on the product bundle @xmath29 as follows : @xmath30 furthermore , the quotient space @xmath31 denoted by @xmath32 recovers @xmath10 up to equivariant homeomorphism .
geometrically , @xmath32 is exactly obtained by gluing @xmath33 copies of @xmath7 along their boundaries by using @xmath6-coloring @xmath13 .
this reconstruction of small covers provides a way of studying closed manifolds by using @xmath6-colored polytopes .
.2 cm in @xcite , izmestiev studied a class of 3-dimensional small covers such that each @xmath13 of @xmath0-colorings on their orbit spaces is 3-colorable ( i.e. , the image of @xmath13 contains only three linearly independent elements of @xmath0 ) .
such a class of small covers are pullbacks from the linear model " in the terminology of davis and januszkiewicz .
izmestiev obtained a classification result , saying that every such small cover can be produced from finitely many 3-dimensional tori with the canonical @xmath0-action under the equivariant connected sum and the equivariant dehn surgery .
.1 cm in this paper , we shall consider all possible 3-dimensional small covers .
our objective is to determine the ( equivariant ) topological types of such a class of 3-dimensional manifolds .
four color theorem guarantees that each simple convex 3-polytope always admits a @xmath0-coloring .
thus , by the reconstruction of small covers , simple convex 3-polytopes with @xmath0-colorings can recover all 3-dimensional small covers , so all simple convex 3-polytopes will be involved in studying 3-dimensional small covers . throughout this paper
, we use the convention that if two simple convex polytopes @xmath34 and @xmath35 are combinatorially equivalent , then @xmath34 is identified with @xmath35 .
.2 cm let @xmath36 denote the set of all pairs @xmath37 where @xmath38 is a simple convex 3-polytope and @xmath13 is a @xmath0-coloring on it , and let @xmath39 denote the set of all 3-dimensional small covers .
then , there is a one - to - one correspondence between @xmath36 and @xmath39 by mapping @xmath37 to @xmath40 .
there is a natural action of @xmath41 on @xmath36 , defined by the correspondence @xmath42 where @xmath43 . obviously , this action is free , and it also induces an action of @xmath41 on @xmath39 by mapping @xmath44 to @xmath45 . both @xmath46 and @xmath45 are @xmath47-equivariantly homeomorphic ( cf @xcite ) , so they are homeomorphic by forgetting their @xmath0-actions .
all elements of each equivalence class of @xmath48 ( resp .
@xmath49 ) are said to be _
.1 cm we shall first carry out our work on @xmath36 .
we shall introduce six operations @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 on @xmath36 . then , under these six operations , up to @xmath56-equivalence we find five basic pairs @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 of @xmath36 where @xmath62 is a 3-simplex , @xmath63 is a 3-sided prism , and @xmath64 , are shown as in the following figure : @xmath65 where @xmath66 is the standard basis of @xmath0 . then the combinatorial version of our main result is stated as follows . [ t1 ] each pair @xmath37 in @xmath36 is an expression of @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , under six operations @xmath73 , @xmath74 , @xmath52,@xmath53 , @xmath54 , @xmath55 . by the reconstruction of small covers , six operations @xmath75 , @xmath53 , @xmath54 , @xmath55 on @xmath36
naturally correspond to six equivariant operations on @xmath39 , denoted by @xmath76 , respectively .
these six operations can be understood very well because of their combinatorial natures .
we shall see that @xmath77 is the equivariant connected sum , and @xmath78 is the equivariant dehn surgery , and other four operations @xmath79 can be understood as the generalized equivariant connected sums . by theorem [ t1 ] , @xmath80 and @xmath81 , @xmath72 , give all elementary generators of the algebraic system @xmath82 . on the other hand , we shall show that @xmath83 is equivariantly homeomorphic to the @xmath1 with canonical linear @xmath0-action , and @xmath84 , are equivariantly homeomorphic to the @xmath2 with four different @xmath0-actions respectively .
then the topological version of our main result is stated as follows .
[ t2 ] each 3-dimensional small cover can be obtained from @xmath1 and @xmath85 with certain @xmath0-actions by using six operations @xmath76 .
theorem [ t2 ] tells us how to obtain a 3-dimensional small cover from only two known small covers @xmath1 and @xmath86 with certain actions by using cut and paste strategies in the sense of six equivariant operations .
this is an equivariant analogue of a well - known result ( @xcite , @xcite , see also @xcite and @xcite ) as follows : `` each closed 3-manifold can be obtained from a 3-sphere @xmath87 or a @xmath87 with one non - orientable bundle by using a finite number of dehn surgeries '' .
as an application , we study the @xmath0-equivariant unoriented cobordism classification of all 3-dimensional small covers .
let @xmath88 denote the set of @xmath0-equivariant unoriented cobordism classes of all 3-dimensional small covers .
then @xmath88 forms an abelian group under disjoint union , so it is also a vector space over @xmath89 .
[ t3 ] @xmath88 is generated by classes of @xmath1 and @xmath85 with certain @xmath0-actions .
it should be pointed out that theorem [ t3 ] is a direct corollary of main theorems in @xcite , but here we shall give it a different proof . actually , the first author of this paper in @xcite dealt with general closed 3-manifolds with effective @xmath90-actions .
let @xmath91 be the @xmath89-vector space consisting of @xmath0-equivariant unoriented cobordism classes of all closed 3-manifolds with effective @xmath90-actions .
then it was shown in @xcite that @xmath91 can be generated by classes of @xmath1 and @xmath85 with certain @xmath0-actions , and each class of @xmath91 contains a small cover as its representative .
in particular , it was also shown in @xcite that @xmath91 has dimension 13 .
thus , @xmath88 has dimension 13 , too .
this paper is organized as follows . in section [ sec2 ]
we establish the six operations on @xmath36 , and then we prove theorem [ t1 ] in section [ sec3 ] . in section [ sec4 ]
we study elementary colored 3-polytopes and determine their equivariant topological types .
moreover , theorem [ t2 ] is settled . in section [ sec5 ]
we discuss how the corresponding six equivariant operations work on @xmath39 . as an application
, we consider the @xmath0-equivariant unoriented cobordism classification of all 3-dimensional small covers and prove theorem [ t3 ] in section [ sec6 ] .
the task of this section is to define six operations on @xmath36 . throughout the remaining part of this paper , each nonzero element of @xmath0
is called a _ color _ , so @xmath0 contains seven colors .
.1 cm first , let us look at all simple uncolored 3-polytopes .
it is well - known that any simple convex 3-polytope can be obtained from a 3-simplex by using three types of _ excision _ methods illustrated in the following figure : cutting out ( i ) a vertex ; ( ii ) an edge ; ( iii ) two edges with a common vertex .
see grnbaum s
book @xcite .
@xmath92 @xmath93 @xmath94 since we shall carry out our study on colored polytopes and small covers , although these three types of excisions are very simple , they can not directly work on colored polytopes and small covers because they will destroy the closedness of small covers .
however , for our purpose we can interpret them as the connected sumoperations with some standard simple @xmath95-polytopes as follows . 1 .
the operation @xmath73 with a 3-simplex @xmath62 @xmath96 2 . the operation @xmath74 with a 3-sided prism @xmath63 @xmath97 3 .
the operation @xmath52 with a truncated prism @xmath98 @xmath99 obviously , each of three operations is invertible as long as we do nt perform the corresponding inverse operations of @xmath100 on @xmath62 , @xmath63 , @xmath98 , respectively .
also , we always can do the operation @xmath73 between any two simple 3-polytopes .
since a 3-sided prism and a truncated prism can be obtained from a 3-simplex by using the operation @xmath73 , we have [ non - coloring ] each simple 3-polytope can be obtained from a 3-simplex under three operations @xmath73 , @xmath74 and @xmath52 .
let @xmath38 be a simple 3-polytope , and let @xmath101 be a facet of @xmath38 .
then @xmath101 is a @xmath102-polygon where @xmath103 .
if @xmath104 , then @xmath101 is called a _ small facet _ ; otherwise , it is called a _
big facet_. also , for two edges with a common vertex in a simple 3-polytope @xmath38 , denoted by @xmath105 , there are at least four neighboring facets around @xmath105 .
but it is easy to see that there are exactly five neighboring facets around @xmath105 if @xmath105 is not in a triangular facet .
since @xmath105 is always associated with the operation @xmath52 , throughout the rest of the paper , we use the _ convention _ that @xmath105 must be chosen in an @xmath106-polygonal facet with @xmath107 , so there are five neighboring facets around @xmath105
. .2 cm suppose that @xmath38 is a simple 3-polytope but it is not a 3-simplex .
then we know by proposition [ non - coloring ] that @xmath38 comes from applying one of the three types of cutting operations on some simple 3-polytope @xmath108 such that the number of facets of @xmath108 is one less than that of @xmath38 . in other words , there is a small facet @xmath101 of @xmath38 such that @xmath108 is obtained by compressing @xmath101 into a point , or an edge or a @xmath105 in @xmath38 . in this case
, we say that @xmath38 is _ compressible at @xmath101 _ , and @xmath108 is the _ compression of @xmath38 at @xmath101_. suppose that @xmath38 is a simple 3-polytope other than a 3-simplex .
then @xmath38 is always compressible at some small facet .
.2 cm now let us carry out our work on @xmath36 .
we wish to know how the three operations @xmath73 , @xmath74 and @xmath52 work on @xmath36 . to make our language more concise ,
first let us give some notions .
[ local colorings ] given a pair @xmath109 in @xmath36 .
let @xmath110 be a vertex ( or a 0-face ) of @xmath38 .
the colors of three facets meeting at @xmath110 are said to be a _ coloring of @xmath110_. let @xmath111 be an edge ( or a 1-face ) of @xmath38 . then there must be four neighboring facets around @xmath111
since @xmath38 is simple , and the colors of these four facets are said to be a _ coloring of @xmath111_. similarly , for a @xmath105 in @xmath38 , there are five neighboring facets around @xmath105 , and then the colors of those facets are said to be a _ coloring of @xmath105_. [ local coloring ] by the definition of @xmath0-colorings , the colors of neighboring facets around a vertex ( resp . an edge and a @xmath105 ) always can span the whole @xmath0 .
it is easy to see that up to @xmath112-equivalence , a vertex admits a unique coloring , an edge admits four different kinds of colorings , and a @xmath105 admits 16 different kinds of colorings .
we list them as follows : 1 .
colorings of a vertex and an edge @xmath113 2 .
colorings of a @xmath105 @xmath114 @xmath115 given a pair @xmath109 in @xmath36 , and let @xmath101 be a facet of @xmath38 .
@xmath101 is said to be _
2-independent _ if the colors of the neighboring facets around @xmath101 span a 2-dimensional subspace of @xmath0 .
similarly , @xmath101 is said to be _
3-independent _ if the colors of the neighboring facets around @xmath101 span the whole @xmath0 . with the above understood ,
let us look at how the three operations @xmath73 , @xmath74 and @xmath52 work on @xmath36 .
[ operation-1 ] up to @xmath50-equivalence , the first two operations @xmath73 and @xmath74 can operate on any vertex and any edge in a colored simple 3-polytope , respectively , and the third operation @xmath52 can operate on @xmath105 in a colored simple 3-polytope as long as the coloring of @xmath105 does not match any one of eight kinds of colorings shown in the figures ( e)-(g ) of remark [ local coloring](2 ) . let @xmath37 be a pair in @xmath36 .
choose a vertex @xmath110 of @xmath38 , since @xmath110 admits a unique coloring up to @xmath56-equivalence , there is a pair @xmath116 such that some vertex in @xmath62 has the same coloring as @xmath110 , so that we can do the operation @xmath73 between @xmath117 and @xmath118 .
choose an edge @xmath111 of @xmath38 , then we know from remark [ local coloring](1 ) that there are four kinds of colorings of @xmath111 up to @xmath56-equivalence , which agree with those colorings of an edge @xmath119 of @xmath63 , as shown in section 1 , where @xmath119 is not an edge of any triangle facet of @xmath63 .
thus , we can perform the operation @xmath74 on @xmath37 with some pair @xmath120 .
choose a @xmath105 ( i.e. , two edges with a common vertex ) in some facet @xmath101 of @xmath38 ( note that @xmath101 is an @xmath106-polygon with @xmath121 by our convention as before ) .
we know from remark [ local coloring](2 ) that there are 16 kinds of colorings of @xmath105 up to @xmath56-equivalence . however , it is easy to see that the eight kinds of colorings shown in the figures ( e)-(g ) can not be used as colorings of the neighboring facets around a pentagon in a simple 3-polytope by the definition of @xmath0-colorings .
this means that if @xmath105 has such a coloring , we can not perform the operation @xmath52 on @xmath37 at @xmath105 . on the other hand , consider a @xmath122 in a truncated prism as shown in the following figure : @xmath123 obviously , @xmath122 admits those eight kinds of colorings shown in the figures ( a)-(d ) of remark [ local coloring](2 ) , but it admits none of eight kinds of colorings shown in the figures ( e)-(g ) of remark [ local coloring](2 ) .
therefore , @xmath37 can do the operation @xmath52 with a @xmath98 with some @xmath0-coloring if @xmath105 admits a coloring which matches one of the eight kinds of colorings shown in the figures ( a)-(d ) of remark [ local coloring](2 ) .
it should be pointed out that @xmath73 can always operate between any two pairs @xmath124 and @xmath125 in @xmath36 up to @xmath50-equivalence .
in fact , choose two vertices @xmath126 and @xmath127 in @xmath34 and @xmath128 respectively , then @xmath126 and @xmath127 have the same coloring up to @xmath50-equivalence .
thus , by applying an automorphism @xmath129 to @xmath130 , we can change the coloring of @xmath126 into that of @xmath127 , so that we can do the operation @xmath73 between @xmath131 and @xmath125 .
we shall see that @xmath73 exactly agrees with the equivariant connected sum of 3-dimensional small covers . similarly to the uncolored case
, clearly we still can not perform the corresponding inverse operations of @xmath132 on colored @xmath62 , @xmath63,@xmath98 , respectively .
however , by remark [ local coloring ] , it is easy to see that for any small facet @xmath101 of a pair @xmath37 , if @xmath101 is 2-independent , then @xmath37 can not be compressed at @xmath101 .
.2 cm by proposition [ non - coloring ] , a natural question is whether each pair @xmath117 of @xmath36 can be produced only from a 3-simplex with @xmath0-colorings in such three operations .
however , generally the answer is _
no_. for example , none of the four colorings on @xmath63 as shown in section 1 can be obtained from a 3-simplex with @xmath0-colorings under three operations @xmath73 , @xmath74 and @xmath52 .
this is because each triangular facet in @xmath63 with any one of those four colorings is 2-independent and it can not be compressed into a point .
more generally , we can further ask the following question : 1 .
_ can any pair @xmath117 be produced by a @xmath95-simplex , a prism and a truncated prism with @xmath0-colorings under operations @xmath73 , @xmath74 and @xmath52 ? _
unfortunately , the answer is still _
no_. actually , generally it is possible that all the small facets are 2-independent , so we can not do the compression of @xmath37 at any of its small facets at all .
this can be seen from the following example .
consider two copies of a square with four neighboring 6-polygons , we can glue them into a simple 3-polytope admitting a 3-colorable coloring , as shown in the following figure : @xmath133 however , this 3-colorable example can not be compressed at any facet under operations @xmath73 , @xmath74 and @xmath52 since each coloring on a 3-simplex ( resp .
a 3-sided prism , and a truncated prism ) is not 3-colorable .
generally , when a pair @xmath37 of @xmath36 is 3-colorable , a theorem of izmestiev in @xcite claims that @xmath37 can be obtained from a finite set of 3-colorable cubes by using the equivariant connected sum ( i.e. , the operation @xmath73 ) and the equivariant dehn surgery .
the reason why his work was carried out very well is because the coloring of a @xmath95-colorable polytope is unique up to @xmath50-equivalence , while generally speaking , the set of all colorings given by more than three colors is quite complicated .
according to the work of izmestiev ( @xcite ) , we might need the fourth operation @xmath53 on @xmath36 .
this operation originally comes from the dehn surgery on 3-manifolds rather than combinatorics . based upon the topological meaning of dehn surgery
, izmestiev gave it a combinatorial description by deleting a quarter of a cylinder with a subsequent gluing of a half - cylinder .
although this combinatorial description of the operation @xmath53 works on @xmath36 very well , it does nt meet the style of this paper , that is , it does not accord with the descriptions of other operations on @xmath36 in this paper .
for this , we give another combinatorial description of this operation @xmath53 , which is shown as follows : @xmath134 where @xmath135 is a quarter of a 3-ball , whose boundary consists of three 2-polygons , three edges and two vertices .
clearly @xmath136 is not a simple 3-polytope , but it still admits a @xmath0-coloring . note that @xmath136 is actually a nice manifold with corners ( @xcite ) .
obviously , the operation @xmath53 is invertible .
however , generally it may not be closed in @xmath36 because doing the operation @xmath53 on a colored 3-polytope @xmath37 might destroy the 3-connectedness of the 1-skeleton of @xmath38 . in the 3-colorable case
, izmestiev showed that if @xmath53 makes the 1-skeleton of the polytope not @xmath95-connected , then one can find a connected sum somewhere else in the original polytope . in the general case , the argument of izmestiev can be carried out to get a generalized result .
the following is the combinatorial lemma proved by izmestiev in @xcite which will be used later in this paper .
[ @xcite ] if the 1-skeleton of a 3-polytope @xmath137 is disconnected after cutting out three non - adjacent edges , then @xmath137 can be written as @xmath138 , where @xmath139 are 3-polytopes .
in addition , when @xmath137 is simple , so are @xmath140 and @xmath141 . next , given a pair @xmath37 in @xmath36 , suppose that we can do an equivariant dehn surgery on @xmath37 , but this operation destroys the 3-connectedness of the 1-skeleton @xmath142 of @xmath38 .
if @xmath13 is 3-colorable , izmestiev gave a canonical method of finding three non - adjacent edges @xmath143 of @xmath38 such that @xmath144 is disconnected ( see @xcite for the argument in detail ) .
then there are two 3-colorable pairs @xmath124 and @xmath145 such that @xmath146 , as shown in the following figure : @xmath147 in the general case , we can still use the izmestiev s method to find the required three non - adjacent edges @xmath143 such that @xmath144 is disconnected , but there are _ two possible colorings _ up to @xmath56-equivalence for three facets determined by @xmath143 , as shown in the following figure : @xmath148 obviously , the case ( i ) is the same as the 3-colorable case above , so there are two pairs @xmath124 and @xmath149 such that @xmath150 . if the case ( ii ) happens , then there still are two pairs @xmath124 and @xmath149 , but we need to introduce a new operation @xmath54 , so that @xmath37 is equal to the sum of @xmath124 and @xmath145 under this new operation @xmath54 .
.2 cm the _ operation @xmath54 _ is defined as follows : first we cut out a triangular facet of @xmath151 , respectively , and then we glue them together along their triangular sections , as shown in the following figure : @xmath152 notice that the operation @xmath54 is invertible .
it should be pointed out that the operation @xmath54 can also work in the case ( i ) .
.2 cm combining the above argument , we have [ composition ] let @xmath37 be a pair in @xmath36 .
suppose that the 3-connectedness of 1-skeleton of @xmath38 is destroyed after doing an equivariant dehn surgery @xmath53 on @xmath37 .
then there are two pairs @xmath124 and @xmath145 in @xmath36 such that either @xmath150 or @xmath153 . next given a pair @xmath37 in @xmath36 and let @xmath101 be a small facet .
we wish to know 1 . _
can @xmath37 always be compressed at @xmath101 if @xmath101 is 3-independent ?
_ to answer this question ( q2 ) , we need to introduce the following operation .
now let us introduce the sixth operation @xmath55 on @xmath36 .
given a pair @xmath37 in @xmath36 , we can not avoid the occurrence of 2-independent facets in @xmath117 in general , but for our propose we can change their colorings .
let @xmath101 be a 2-independent @xmath18-polygonal facet of @xmath117 .
then we can construct a @xmath18-sided prism @xmath154 $ ] , which naturally admits a coloring @xmath155 such that the coloring of the neighboring facets around the top facet ( or bottom facet ) is the same as that of @xmath101 in @xmath37 . since @xmath101 is 2-independent
, we can give two different colorings on the top facet and the bottom facet of @xmath156 , such that the bottom facet of @xmath156 has the same coloring as @xmath101 .
then we can define an operation between @xmath37 and @xmath157 as follows : cutting out the @xmath101 of @xmath38 and the bottom facet of @xmath156 , and then gluing them together along sections , as shown in the following figure : @xmath158 this operation exactly changes the coloring of @xmath101 , so we also call it the _ coloring change _ , denoted by @xmath55 .
clearly , the operation @xmath55 is invertible .
.2 cm we shall mainly consider the coloring changes of 2-independent small facets , all possible cases ( in the sense of @xmath56-equivalence ) of which are listed as follows : 1
. triangular case @xmath159 2 .
rectangular case @xmath160 3 .
pentagonal case @xmath161 [ section1 ] as seen as above , when we do those six operations on @xmath36 , we need to cut out vertices , edges , @xmath105 s , 2-independent triangular facets , 2-independent square facets , and 2-independent pentagonal facets , so that we can produce different kinds of sections on polytopes .
by @xmath162 , @xmath163 , and @xmath164 we denote those sections obtained by cutting out a vertex @xmath110 , an edge @xmath111 and a @xmath105 , a 2-independent triangular facet , a 2-independent square facet and a 2-independent pentagonal facet respectively .
also , the colorings of neighboring facets around @xmath162 , @xmath165 , and @xmath164 are said to be the _ colorings _ of @xmath162 , @xmath163 , and @xmath164 respectively . obviously , these sections have the properties : 1 .
the colorings of @xmath162 are all 3-independent . up to @xmath50-equivalence
, @xmath166 admits a unique coloring , @xmath167 admits four different colorings , and @xmath168 admits eight different colorings .
the colorings of @xmath166 and @xmath167 agree with the colorings of a vertex and an edge respectively , see remark [ local coloring](1 ) .
the colorings of @xmath168 agree with the colorings shown in the figures ( a)-(d ) of remark [ local coloring](2 ) .
the colorings of @xmath169 are all 2-independent . up to @xmath56-equivalence , @xmath170 and @xmath164
admit a unique coloring , but @xmath171 admits two different colorings . these colorings agree with the colorings of around 2-independent small facets , as shown before remark [ section1 ] .
notice that @xmath166 and @xmath170 are triangular sections , @xmath167 and @xmath171 are rectangular sections , and @xmath168 and @xmath164 are pentagonal sections .
finally , let us discuss the question ( q2 ) .
[ compress ] let @xmath37 be a pair in @xmath36 and let @xmath101 be a small facet of @xmath38 . then @xmath37 is compressible at @xmath101 if and only if @xmath101 is 3-independent . since we can not perform the corresponding inverse operations of @xmath132 on colored @xmath62 , @xmath63,@xmath98 , respectively , we may assume that when @xmath101 is a triangular ( resp .
rectangular , or pentagonal ) facet , @xmath38 is not a 3-simplex ( resp .
a @xmath63 , or a @xmath98 ) . obviously , if @xmath37 is compressible at @xmath101 then @xmath101 is 3-independent .
conversely , our argument proceeds as follows .
.2 cm ( 1 ) suppose that @xmath101 is a 3-independent triangular facet .
then it is easy to see that @xmath37 is compressible at @xmath101 .
.2 cm ( 2 ) suppose that @xmath101 is a 3-independent rectangular facet .
then up to @xmath172-equivalence , we may list all possible colorings of @xmath101 and its four neighboring facets as follows : @xmath173 where @xmath174 with @xmath175 , and at least one of @xmath176 and @xmath177 is nonzero .
obviously , if none of the four neighboring facets around @xmath101 is triangular , then one can always compress @xmath101 into an edge along @xmath178 ( if @xmath179 ) or @xmath180 ( if @xmath181 ) , as shown in the following figure .
@xmath182 if there are triangular neighboring facets around @xmath101 , then by steinitz s theorem the number of such triangular neighboring facets is at most 2 . since @xmath38 is not a @xmath63 by our assumption , the number must be 1 . with no loss assume that @xmath183 is a triangular facet . if @xmath183 is 3-independent , then one first compress @xmath183 into a point , so that @xmath101 becomes a 3-independent triangular facet .
furthermore , one can compress @xmath101 into a point .
if @xmath183 is 2-independent and @xmath184 , then one can compress @xmath101 into an edge along @xmath178 ; if @xmath183 is 2-independent but @xmath185 , then one can first change the coloring of @xmath183 into @xmath186 by the operation @xmath55 , so that one can also compress @xmath101 into an edge along @xmath178 .
.2 cm ( 3 ) suppose that @xmath101 is a 3-independent pentagonal facet
. then up to @xmath187-equivalence , all possible colorings of @xmath101 and its five neighboring facets may be listed as follows : @xmath188 where @xmath189 with @xmath190 and at least one of @xmath191 , and @xmath192 is nonzero .
an easy observation shows that if none of the five neighboring facets around @xmath101 is triangular , then there is always at least one neighboring facet @xmath193 around @xmath101 so that one may compress @xmath101 into a @xmath105 along @xmath193 .
if there are triangular neighboring facets around @xmath101 , then the steinitz s theorem makes sure that the number @xmath102 of such triangular neighboring facets is at most 2 .
.2 cm when the number @xmath102 is just 2 , there are two possibilities . with no loss
assume that either @xmath183 and @xmath178 or @xmath183 and @xmath180 are triangular .
if @xmath183 and @xmath180 are triangular , then it is easy to check that at least one of @xmath183 and @xmath180 is 3-independent , so that one can compress the 3-independent triangular facet into a point and then @xmath101 will become a rectangular facet . thus , the problem is reduced to the case ( 2 ) above .
if @xmath183 and @xmath178 are triangular , then there are two possible cases : either at least one of @xmath183 and @xmath178 is 3-independent or both @xmath183 and @xmath178 are 2-independent . if the former case happens , in a same way as above , one may reduce this case to the case ( 2 ) .
if the latter case happens , by changing the coloring of @xmath183 or @xmath178 via the operation @xmath55 ( if necessary ) , one may compress @xmath101 into a @xmath105 along @xmath183 or @xmath178 .
.2 cm when the number @xmath102 is just 1 , with no loss assume that @xmath183 is triangular .
if @xmath183 is 3-independent , then one may compress @xmath183 into a point , so that the case is reduced to the case ( 2 ) .
if @xmath183 is 2-independent , then an easy argument shows that , by adjusting the coloring of @xmath183 via the operation @xmath55 ( if necessary ) , one may always compress @xmath101 into a @xmath105 along @xmath183 or @xmath178 or @xmath180 .
we see from the proof of proposition [ compress ] that if @xmath101 is rectangular or pentagonal , then the manner of compressing @xmath101 not only depends upon the colorings of the neighboring facets around @xmath101 , but also the existence of the triangular neighboring facets around @xmath101 . in general , the compression at @xmath101 may need two possible steps : first compress the neighboring triangular facets around @xmath101 , and then perform the compression of @xmath101 .
so after the compressions , @xmath101 may be compressed into an edge or a point if @xmath101 is rectangular , and a @xmath105 or an edge or a point if @xmath101 is pentagonal .
let @xmath37 be a pair in @xmath36 .
we shall finish the proof of theorem [ t1 ] by using the descending induction on the number of facets of the simple @xmath95-polytope @xmath38 . without the loss of generality ,
assume that @xmath38 contains big facets .
.2 cm first , by proposition [ compress ] , we can compress all possible 3-independent small facets until we can not find them anymore .
this does not increase the number of facets of @xmath38 .
let @xmath194 be the compression of @xmath37 under this step , and assume that @xmath195 still contains big facets .
then we divide our argument into two cases : 1 .
there are adjacent big facets in @xmath196 ; 2 .
there are no adjacent big facets in @xmath196 .
.2 cm * case ( a ) . *
suppose that there are adjacent big facets in @xmath196
. then there must be a pair of adjacent big facets such that there is an adjacent small facet as shown in the following picture : @xmath197 this is because the facets of @xmath196 are not all big according to the euler characteristic of @xmath198 .
next we try to do the equivariant dehn surgery @xmath53 on @xmath195 .
.2 cm when @xmath199 and @xmath200 have the same coloring , we can do dehn surgery @xmath53 on @xmath201 , which would reduce the number of facets by one . if this operation does nt destroy the 3-connectedness of 1-skeleton of @xmath202 , then we go on with our induction . or else , by proposition [ composition ] we have that @xmath195 can be separated into two smaller pairs @xmath124 and @xmath145 such that either @xmath203 or @xmath204
. then the problem is reduced to carrying out our inductions on @xmath124 and @xmath145 .
.2 cm when @xmath199 and @xmath200 have different colorings , since the local coloring around @xmath200 is @xmath205-independent , by the operation @xmath55 we can change the coloring of @xmath200 to match the coloring of @xmath199
. then we can do the dehn surgery operation , turning back to the above case .
.2 cm the above procedure can always be carried out until we can not find adjacent big facets anywhere .
.2 cm * case ( b ) .
* if there are no adjacent big facets in @xmath196 , then any big facet is surrounded by 2-independent small facets . by the operation @xmath55
, we can change the coloring of a small facet , say @xmath101 , then the adjacent small facets around @xmath101 become @xmath95-independent .
then we can compress them by using operations @xmath73 , @xmath74 and @xmath52 .
we note that the edge number of the big facet will be reduced while we compress its neighboring triangular facets , and this number will be either reduced or unchanged while we compress its neighboring rectangular facets , but this number will be unchanged or reduced or becoming bigger while we compress its neighboring pentagonal facets , as shown in the following figure : @xmath206 in particular , when we compress 3-independent pentagons , it is possible to produce new big facets .
for example , if @xmath183 is a pentagon in the above figure , then it will become a big facet after compressing @xmath207 .
in addition , it is easy to see that compressing rectangles and pentagons may lead to the adjacency of big facets .
if this happens , we can return to the case ( a ) to do dehn surgeries .
otherwise , by changing the colorings of small facets and compressing them , we can carry on our work to reduce the edge numbers of big facets .
.2 cm these alternate processes above can always end in finite steps since the number of facets of @xmath38 is finite . note that with the help of @xmath55 , five operations @xmath73 , @xmath74 , @xmath52 , @xmath53 , @xmath54 can not only decrease the number of facets , but they can also decrease the edge numbers of big facets .
.2 cm on the other hand , we see that the six operations themselves also involve some special colored blocks . specifically , when doing three operations @xmath73 , @xmath74 and @xmath52 , the colored @xmath62 , @xmath63 and @xmath98 are involved ;
when doing the operation @xmath53 , the pair @xmath208 is involved . as seen above , using the equivariant dehn surgery @xmath53 , we can avoid changing the colorings of big facets .
so the special colored blocks involved in the operation @xmath55 are only those colored @xmath209-sided prisms @xmath210 with top and bottom facets differently colored and being both 2-independent where @xmath211 .
.2 cm combining cases ( a ) and ( b ) , we can always finish our induction by using the six types of operations until we reach one of those colored 3-polytopes @xmath62 , @xmath98 , and @xmath210 @xmath212 above .
furthermore , by reversing the induction process , eventually @xmath37 can be described as an expression of those colored blocks @xmath62 , @xmath98 , @xmath210 @xmath212 , and @xmath135 under the six operations . .2 cm next , to complete the proof , let us make a further analysis for those colored blocks @xmath62 , @xmath98 , @xmath210 @xmath212 , and @xmath135
. .2 cm first it is easy to see that @xmath62 and @xmath213 admit a unique @xmath0-coloring up to @xmath50-equivalence , as shown in the following figure : @xmath214 in particular , we have that @xmath215 , and this procedure is shown as follows : @xmath216 we know from ( * ? ? ?
* theorem 3.1 ) that @xmath63 admits five kinds of colorings up to @xmath56-equivalence , which are listed as follows : @xmath217 obviously , the colored @xmath95-sided prism on the right is the connected sum of two colored 3-simplices , as shown above .
now let us show that a colored @xmath98 or @xmath218 or @xmath219 can be obtained from colored 3-simplices and 3-sided prisms via only two operations @xmath73 and @xmath74 . 1 . from figures ( a)-(d ) of remark [ local coloring](2 )
, we can obtain that @xmath220 admits nine kinds of colorings up to @xmath50-equivalence , as listed in the following figure : @xmath221 we claim that up to @xmath50-equivalence , by doing the operation @xmath73 of colored @xmath63 s with colored @xmath62 s , we can obtain the required nine kinds of colorings on @xmath98 .
the argument is not difficult . see the following figure for two special cases , and all the other cases are similar to these two .
notice that the connected sums @xmath73 of a colored @xmath62 with some vertex of the top facet of a colored @xmath63 and with some vertex of the bottom facet of a colored @xmath63 respectively may produce different colorings of @xmath98 .
@xmath222 2 . consider the colored prisms @xmath218 and @xmath219 with 2-independent top and bottom facets differently colored .
it is easy to see that up to @xmath50-equivalence and an automorphism @xmath11 of @xmath210 , @xmath218 admits six such colorings , and @xmath219 admits three such colorings , where @xmath11 is the automorphism of rotating facets on the side . we list them as follows : @xmath223 note that clearly @xmath11 has no influence on the reconstruction of the above colored 3-polytopes up to equivariant homeomorphisms ( cf @xcite ) . similarly to the case ( a ) , an easy argument shows that each of colored 4-sided prisms shown in figures ( p ) and ( q ) is the sum of two colored 3-sided prisms under the operation @xmath74 , and each of colored 5-sided prisms shown in figure ( r ) is the sum of a colored 3-sided prism and a colored 4-sided prism under the operation @xmath74 so it is also the sum of three colored 3-sided prisms under the operation @xmath74 .
.2 cm with all above arguments together , we see that , up to @xmath56-equivalence there are only five elementary colored 3-polytopes as stated in section 1 , which can produce all colored 3-polytopes under the six operations .
this completes the proof of theorem [ t1 ] . @xmath224 [ prism ] for a colored @xmath106-sided prism @xmath225 , since its facets on the side are all squares , by considering 2-independence and 3-independence of square facets , we can use operations @xmath74 and @xmath55 alternately to compress facets on the side , so that @xmath225 can be obtained from the colored @xmath63 and @xmath218 . since each colored 4-sided prism used in the operation @xmath55 above can be expressed as the sum of two colored @xmath63 s under the operation @xmath74 by the proof of theorem [ t1 ] , we conclude that @xmath225 is a sum of some colored @xmath63 s under the operations @xmath74 and @xmath55 .
the main task of this section is to determine those 3-dimensional small covers corresponding to @xmath226 and @xmath227 , as stated in section 1 .
.2 cm recall that a small cover @xmath228 is equivariantly homeomorphic to its reconstruction @xmath229 where the pair @xmath230 is determined by @xmath231 .
it is well - known ( see @xcite and @xcite ) that @xmath5-dimensional real projective space @xmath232 admits a canonical linear @xmath6-action defined by @xmath233\longmapsto [ x_0,g_1x_1, ... ,g_nx_n]\ ] ] where @xmath234 . this action fixes @xmath235 fixed points @xmath236 $ ] , @xmath237 , and its orbit space is homeomorphic to the image of the map @xmath238 by @xmath239)=\big({{|x_0|}\over{\sum_{i=0}^n|x_i| } } , { { |x_1|}\over{\sum_{i=0}^n|x_i| } } , ... , { { |x_n|}\over{\sum_{i=0}^n|x_i|}}\big).\ ] ] it is easy to see that the image of @xmath240 is an @xmath5-dimensional simplex .
a direct observation shows that the @xmath235 facets of this @xmath5-simplex are colored by @xmath241 respectively , where @xmath242 is the standard basis of @xmath6 .
this gives [ basic small-1 ] @xmath243 is equivariantly homeomorphic to the @xmath1 with a canonical linear @xmath0-action .
the product of @xmath244 and @xmath245 with canonical linear actions gives a canonical @xmath0-action ( denoted by @xmath246 ) on @xmath2 , which has exactly six fixed points .
explicitly , this action on the product @xmath2 is defined by @xmath247\big)\big)\longmapsto \big((x_0,g_1x_1 ) , [ y_0 , g_2y_1,g_3y_2]\big).\ ] ] the orbit space of this action on @xmath2 is the product of a 1-simplex and a 2-simplex , so it is just a 3-sided prism .
it is also easy to see that the orbit space of this action admits the same coloring as @xmath248 .
thus we have [ basic small-2 ] @xmath249 is equivariantly homeomorphic to the product @xmath2 with the canonical linear action @xmath246 .
regard @xmath250 as the unit circle @xmath251 in @xmath252 and @xmath253 as the projective plane @xmath254\big| v\in { \mathbb c } , w\in { \mathbb r}\ , \}$ ] in @xmath255 , we then construct three @xmath0-actions @xmath256 on @xmath85 as follows : 1 .
the action @xmath257 on @xmath2 is defined by the following three commutative involutions @xmath258 ) & \longmapsto ( \bar{z } , [ zv , w ] ) \\
t_2 : ( z,[v , w ] ) & \longmapsto ( z , [ -\bar{z}\bar{v},w])\\ t_3 : ( z,[v , w ] ) & \longmapsto ( \bar{z } , [ -zv , w ] ) . \end{aligned}\ ] ] 2 .
the action @xmath259 on @xmath2 is defined by the following three commutative involutions @xmath258 ) & \longmapsto ( \bar{z } , [ zv , w ] ) \\ t_2 : ( z,[v , w ] ) & \longmapsto ( z , [ \bar{z}\bar{v},w])\\ t_3 : ( z,[v , w ] ) & \longmapsto ( \bar{z } , [ -zv , w ] ) . \end{aligned}\ ] ] 3 .
the action @xmath260 on @xmath2 is defined by the following three commutative involutions @xmath258 ) & \longmapsto ( \bar{z } , [ \bar{z}v , w ] ) \\ t_2 : ( z,[v , w ] ) & \longmapsto ( z , [ z\bar{v},w])\\ t_3 : ( z,[v , w ] ) & \longmapsto ( \bar{z } , [ -z\bar{v},w ] ) . \end{aligned}\ ] ] note that the action @xmath260 was first given in @xcite .
these three actions fix the same six points @xmath261)$ ] , @xmath262)$ ] and @xmath263)$ ] , where @xmath264 .
[ basic small-3 ] @xmath265 are equivariantly homeomorphic to @xmath266 respectively . first , let us show that each orbit space of the three actions is homeomorphic to a 3-sided prism @xmath63 . for @xmath267 and @xmath268 , write @xmath269 and @xmath270 where @xmath271 $ ] , @xmath272 , and @xmath273 $ ] .
then we define the map @xmath274 by @xmath275)=(x_1 , x_2 , x_3 , x_4 , x_5)\ ] ] where @xmath276 notice that @xmath277=\cos(2\pi t-\theta)$ ] and @xmath278|=|\sin(2\pi t-\theta)|$ ] .
obviously , this map @xmath240 is compatible with three actions @xmath279 on @xmath85 .
in particular , we easily see that for each @xmath271 $ ] , the image of @xmath240 restricted to @xmath245 is a 2-simplex , which consists of all triples @xmath280 .
also , the set @xmath281\}$ ] forms a 1-simplex .
thus , the image of @xmath240 is a 3-sided prism .
furthermore , it is easy to see that each orbit space of the three actions is homeomorphic to this 3-sided prism .
.2 cm next we show that the orbit space of the action @xmath282 admits the same coloring as @xmath283 .
we shall only consider the case @xmath284 because the arguments of other two cases are similar .
our strategy is to first determine the tangent representations at those fixed points and then to give the coloring on the orbit space by using algebraic duality .
.2 cm @xmath285 , which consists all homomorphism from @xmath0 to @xmath286 , gives all irreducible representations of @xmath0 , and forms an abelian group with addition given by @xmath287 , where @xmath288 .
the homomorphisms @xmath289 form a basis of @xmath290 . now write @xmath291 . when @xmath292 , the action @xmath257 restricted to @xmath293 can be defined by the following way @xmath294)\big)\longmapsto & \big(-1 , [ \rho_1(g)\rho_2(g)v_1 , \rho_1(g)\rho_2(g)\rho_3(g)v_2 , w]\big)\\ = & \big(-1 , [ \rho_3(g)v_1 , v_2 , \rho_1(g)\rho_2(g)\rho_3(g)w]\big)\\ = & \big(-1 , [ v_1 , \rho_3(g)v_2 , \rho_1(g)\rho_2(g)w]\big)\end{aligned}\ ] ] and when @xmath295 , the action @xmath257 restricted to @xmath296 can be defined by the following way @xmath297)\big)\longmapsto & \big(1 , [ \rho_2(g)v_1 , \rho_2(g)\rho_3(g)v_2 , w]\big)\\ = & \big(1 , [ \rho_3(g)v_1 , v_2 , \rho_2(g)\rho_3(g)w]\big)\\ = & \big(1 , [ v_1 , \rho_3(g)v_2 , \rho_2(g)w]\big)\end{aligned}\ ] ] then we can read off the tangent representations at six fixed points , which determine a @xmath285-coloring on 1-skeleton of the orbit space by gkm theory ( see @xcite and @xcite ) , as shown in the following figure : @xmath298 this @xmath285-coloring is dual to the @xmath0-coloring on the orbit space by @xmath299 ( cf ) , so we can obtain the desired coloring , as shown in the above figure .
although @xmath135 is not a 3-polytope , it is contractible , so we can apply the method of reconstruction of small covers to @xmath300 to obtain a 3-manifold , denoted by @xmath301 .
[ sphere ] @xmath302 is equivariantly homeomorphic to the @xmath87 with the standard @xmath0-action .
moreover , so is @xmath303 . consider the standard @xmath0-action on @xmath87 by @xmath304 obviously , this action has two fixed points @xmath305 , and its orbit space is identified with @xmath135 .
a direct observation shows that three 2-polygon faces of the orbit space are colored by @xmath306 , so this agrees with the coloring @xmath155 on @xmath135 .
since @xmath213 is contractible , any principal @xmath0-bundle over @xmath135 is trivial .
furthermore , by the method of reconstruction it is easy to see that @xmath307 is equivariantly homeomorphic to the @xmath87 with the standard @xmath0-action .
we easily see from ( * ? ? ?
* theorem 3.1 ) that @xmath308 is not a small cover .
in fact , any @xmath5-sphere @xmath309 with @xmath310 can not become a small cover .
this is because its mod 2 betti numbers @xmath311 can not be used as the @xmath11-vector of any simple convex @xmath5-polytope .
but @xmath250 is a small cover . also , it is easy to see that both @xmath308 and @xmath312 are @xmath47-equivariantly homeomorphic , where @xmath313 . by the reconstruction of small covers , together with theorem [ t1 ] and lemmas [ basic small-1 ] , [ basic small-2 ] , [ basic small-3 ] and [ sphere ] ,
we have completed the proof of theorem [ t2 ] .
it remains to understand the geometrical meanings of corresponding six operations on @xmath39 .
now let us look at how corresponding six operations work on @xmath39 .
in particular , this will tell us how to construct a small cover 3-manifold by using cut and paste strategies .
.2 cm to understand six operations on @xmath39 , first let us study the corresponding geometrical meanings of sections @xmath162 , @xmath314 in small covers .
these sections actually correspond to some closed surfaces , which we list in the following lemma .
[ section ] the corresponding geometrical meanings ( up to homeomorphism ) of sections @xmath162 , @xmath169 in small covers are stated as follows : 1 .
@xmath166 corresponds to a 2-sphere @xmath315 ; 2 .
@xmath167 corresponds to a 2-dimensional torus @xmath316 or a klein bottle @xmath317 shown as follows :
@xmath318 3 .
@xmath168 corresponds to a @xmath319 or a @xmath320 shown as follows : @xmath321 4 .
@xmath170 corresponds to a disjoint union @xmath322 ; 5 .
@xmath171 corresponds to a @xmath323 or a @xmath324 shown as follows : @xmath325 6 .
@xmath164 corresponds to a disjoint union @xmath326 , where @xmath327 denotes the ordinary connected sum .
the argument is not quite difficult , and it is mainly based upon the reconstruction method of small covers .
we would like to leave it to readers as an exercise .
lemma [ section ] will play a beneficial role in understanding the six operations on @xmath39 .
it should be pointed out that the corresponding closed surfaces of those sections are all not small covers in the sense of davis - januszkiewicz .
actually , for each such section @xmath207 , its corresponding closed 2-manifold @xmath328 is the double covering space of a small cover over @xmath207 .
also , if @xmath207 is 2-independent then @xmath328 is disconnected , and if @xmath207 is 3-independent then @xmath328 is connected .
this operation is actually the equivariant connected sum . by lemma [ section ] , cutting out a vertex @xmath110 of a colored @xmath37 exactly corresponds to cutting out a @xmath0-invariant open 3-ball which contains a fixed point of @xmath46 as shown in the following figure , so that the operation @xmath73 on @xmath36 induces the equivariant connected sum @xmath77 on @xmath39 .
@xmath329 now from the proof of theorem [ t1 ] and by lemmas [ basic small-1 ] , [ basic small-2 ] , [ basic small-3 ] and [ section ] , we have [ top - type1 ] the topological type of @xmath330 is either @xmath331 or @xmath86 .
furthermore , the topological type of @xmath332 is either @xmath333 or @xmath334 .
by lemma [ section ] , when we do the operation @xmath335 on a @xmath46 , we exactly cut out a @xmath0-invariant open solid torus @xmath336 ( or a @xmath0-invariant open solid klein bottle @xmath337 ) from @xmath40 , while we also need to cut out a same type of @xmath0-invariant open solid torus ( or a same type of @xmath0-invariant open solid klein bottle ) from a @xmath330 .
however , by corollary [ top - type1 ] @xmath338 has two different topological types : either @xmath331 or @xmath85 . according to the colorings on @xmath63 ,
an easy argument shows that when the topological type of @xmath338 is @xmath331 , we can only cut out a @xmath0-invariant open solid torus from @xmath338 , but when the topological type of @xmath338 is @xmath85 , we can not only cut out a @xmath0-invariant open solid torus but also a @xmath0-invariant open solid klein bottle from @xmath338 .
more precisely , up to @xmath56-equivalence , when @xmath339 , we can only cut out a @xmath0-invariant open solid torus from @xmath340 and when @xmath341 , we can only cut out a @xmath0-invariant open solid klein bottle from @xmath340 .
thus , we have that if the topological type of @xmath338 is @xmath331 , then @xmath342 and if the topological type of @xmath338 is @xmath85 , then @xmath343 similarly , by lemma [ section ] , when we do the operation @xmath344 on a @xmath46 , we need to cut out a same type of @xmath0-invariant @xmath345 ( or a same type of @xmath0-invariant @xmath346 ) from @xmath40 and @xmath347 respectively , and then glue the remaining parts together along their boundaries , where @xmath345 ( resp .
@xmath346 ) denotes the interior of a 3-dimensional @xmath0-manifold with boundary @xmath319 ( resp .
@xmath320 ) . we know from corollary [ top - type1 ] that the topological type of @xmath347 is either @xmath333 or @xmath348 . according to the colorings on @xmath98 , we see easily that if the topological type of @xmath347 is @xmath333 , then we can only cut out a @xmath0-invariant @xmath345 from @xmath347 , and if the topological type of @xmath347 is @xmath348 , we can only cut out a @xmath0-invariant @xmath346 from @xmath347
. therefore , we have that when the topological type of @xmath347 is @xmath333 , @xmath349 and when the topological type of @xmath347 is @xmath348 , @xmath350 recall ( cf @xcite and @xcite ) that a @xmath351-type dehn surgery on a 3-manifold @xmath352 is as follows : removing a solid torus from @xmath352 and then sewing it back in @xmath352 such that the meridian goes to @xmath353 times the longitude and @xmath354 times the meridian , where @xmath355 . .2 cm _ claim . _
the operation @xmath78 on @xmath46 is exactly an equivariant @xmath356-type dehn surgery on @xmath40 .
.2 cm in fact , when we cut out an edge from @xmath300 , the section is a 3-colorable square , so by lemma [ section ] we exactly cut out a @xmath0-invariant open solid torus from @xmath301 . on the other hand , using the method of the reconstruction of small covers , the remaining part of the @xmath357 can be reconstructed into a @xmath0-invariant solid torus .
so the operation @xmath78 will remove a @xmath0-invariant open solid torus @xmath358 from @xmath40 and glue back another @xmath0-invariant solid torus @xmath359 come from @xmath301 , mapping the meridian ( longitude ) of @xmath359 to the longitude ( meridian ) of @xmath358 .
notice that each edge in @xmath109 corresponds to a circle in @xmath40 by the reconstruction of small covers .
.2 cm therefore , the operation @xmath78 on a @xmath44 up to @xmath50-equivalence can be expressed as follows : @xmath360 when we do the operation @xmath361 on two @xmath362 and @xmath363 , since @xmath170 corresponds to a disjoint union @xmath364 by lemma [ section ] , we need to cut out a @xmath0-invariant @xmath365 from each of both @xmath362 and @xmath363 .
then we glue them together along their boundaries .
thus , we have @xmath366 notice that the two @xmath0-invariant @xmath365 cut out from @xmath362 and @xmath363 may not always be equivariantly homeomorphic because we may cut out two triangular facets with different colorings from @xmath124 and @xmath145 . as we have seen , when we do the operation @xmath55 on @xmath36 , only 2-independent small facets are involved .
thus , when we do the operation @xmath367 on a @xmath40 , by lemma [ section](4)-(6 ) we need to cut out a @xmath0-invariant @xmath365 , or a @xmath0-invariant @xmath368 , or a @xmath0-invariant @xmath369 , or a @xmath0-invariant @xmath370 from @xmath40 , and at the same time , up to @xmath56-equivalence we also need to do same things on @xmath371 , where the top facet and the bottom facet of each @xmath372 are colored by two different colors and the colorings of neighboring facets around them are 2-independent , as shown in figures ( h),(p)-(r ) of section [ sec3 ] , then gluing their corresponding boundaries together .
when @xmath373 , by lemmas [ basic small-2 ] and [ basic small-3 ] , the topological type of @xmath338 is exactly @xmath2 .
when @xmath374 , we know from the proof of theorem [ t1 ] that @xmath375 is the sum of two @xmath376 and @xmath377 under @xmath378 , and @xmath379 is the sum of a @xmath380 and a @xmath381 under @xmath378 .
however , this does not make clear what the topological types of @xmath375 and @xmath382 are .
next , we shall investigate their topological types .
.2 cm it is well known ( see @xcite ) that for any closed surface @xmath383 , @xmath383-bundles over @xmath250 are classified by the mapping class group @xmath384 . in particular , 1 .
when @xmath383 is a torus @xmath316 , @xmath385 .
when @xmath383 is a klein bottle @xmath317 , @xmath386 .
in fact , if we think of @xmath317 as @xmath387 , then elements in @xmath388 can be represented by @xmath389 where @xmath390 ) = ( [ z_1^{\epsilon_1 } , z_2^{\epsilon_2}])$ ] .
first let us look at the three colored 4-sided prisms shown in figure ( p ) of section [ sec3 ] , denoted by @xmath391 , respectively .
[ 4-prism1 ] @xmath392 , are equivariantly homeomorphic to three twisted @xmath316-bundles over @xmath250 with monodromy maps @xmath393 , @xmath394 , @xmath395 , respectively , where the @xmath0-action on each twisted @xmath316-bundle over @xmath250 is induced by the @xmath0-action @xmath396 on @xmath397 $ ] defined by the following three commutative involutions @xmath398 by lemma [ section ] , any horizontal section of each @xmath399 corresponds to a disjoint union @xmath323 in @xmath400 .
this means that the two parts obtained by cutting each @xmath401 horizontally correspond to two @xmath0-invariant @xmath316-handlebodies @xmath316-@xmath402 and @xmath316-@xmath403 , each of which is homeomorphic to @xmath397 $ ] , as shown in the following figure : @xmath404 obviously , all @xmath316-@xmath402 s are equivariantly homeomorphic to the @xmath397 $ ] with the @xmath0-action @xmath396 .
an easy observation shows that @xmath316-@xmath405 , are obtained from the @xmath397 $ ] with the @xmath0-action @xmath396 by using the following dehn twists on @xmath397 $ ] @xmath406 respectively .
namely , the topological types of @xmath316-@xmath407 are @xmath408)=\ { ( e^{\pi ( t+1)\text{\bf i}}z_1,z_2,t)\big| z_1 , z_2\in s^1 , t\in [ -1,1]\}\\ & d_2(t\times[-1,1])=\ { ( z_1,e^{\pi ( t+1)\text{\bf i}}z_2,t)\big| z_1 , z_2\in s^1 , t\in [ -1,1]\}\\ & d_3(t\times[-1,1])=\ { ( e^{\pi ( t+1)\text{\bf i}}z_1,e^{\pi ( t+1)\text{\bf i}}z_2,t)\big| z_1 , z_2\in s^1 , t\in [ -1,1]\ } \end{aligned}\ ] ] respectively , and they admit the @xmath0-actions which are compatible with the @xmath0-action @xmath396 on @xmath397 $ ] , as follows : 1 .
the @xmath0-action @xmath409 on @xmath410)$ ] is given by the following three commutative involutions @xmath411 satisfying @xmath412 .
2 . the @xmath0-action @xmath413 on @xmath414)$ ]
is given by the following three commutative involutions @xmath415 satisfying @xmath416 .
3 . the @xmath0-action @xmath417 on @xmath418)$ ]
is given by the following three commutative involutions @xmath419 satisfying @xmath420 .
when @xmath421 , we have @xmath422 , so we see that each @xmath423 is obtained by equivariantly gluing @xmath424 $ ] and @xmath425)$ ] along their boundaries via the identity of @xmath316 . on the other hand , when @xmath426 , we have @xmath427 , so we see that the three dehn twists @xmath428 determine exactly three monodromy maps @xmath429 , as follows : @xmath430 this completes the proof .
let @xmath431 , denote those three colored 4-sided prisms shown in figure ( q ) of section [ sec3 ] . in a similar way , we can prove the following lemma .
[ 4-prism2 ] @xmath432 , are equivariantly homeomorphic to three twisted @xmath317-bundles over @xmath250 with monodromy maps @xmath433 and @xmath434 respectively , where the @xmath0-action on each twisted @xmath317-bundle over @xmath250 is induced by the @xmath0-action @xmath435 on @xmath436 $ ] defined by the following three commutative involutions @xmath437 , t)\longmapsto ( [ \bar{z}_1,z_2],t)\\ & t_2:([z_1,z_2 ] , t)\longmapsto ( [ z_1,\bar{z}_2],t)\\ & t_3:([z_1,z_2 ] , t)\longmapsto ( [ z_1,z_2],-t ) . \end{aligned}\ ] ] let @xmath438 where @xmath439 is a punctured torus and @xmath440 is a mbius band with @xmath441 . then @xmath442 is homeomorphic to @xmath443 .
it is well known ( see @xcite ) that any diffeomorphism of @xmath442 is isotopic to one leaving @xmath439 and @xmath440 invariant , and there is the following result .
the extended mapping class group @xmath444 of @xmath442 is isomorphic to @xmath445 , and the isomorphism is given by the natural homomorphism @xmath446 where @xmath447 is a torus .
let @xmath448 , denote those three colored 5-sided prisms shown in figure ( r ) of section [ sec3 ] .
then we have [ 5-prism ] @xmath449 , are equivariantly homeomorphic to three special twisted @xmath442-bundles over @xmath250 with monodromy maps as the inverse images of @xmath450 , @xmath394 , @xmath451 respectively under the isomorphism @xmath452 .
in fact , each @xmath453 can be constructed by using a colored 3-sided prism and a colored 4-sided prism under the operation @xmath74 , as shown in the following figure : @xmath454 by lemmas [ basic small-2 ] and [ basic small-3 ] , each colored 3-sided prism used above corresponds to a trivial @xmath253-bundle over @xmath250 , and by lemma [ 4-prism1 ] the three colored 4-sided prisms used above correspond to the three nontrivial @xmath316-bundles over @xmath250 with monodromy matrices @xmath450 , @xmath394 , @xmath455 , respectively .
so each @xmath456 is equivariantly homeomorphic to a non - trivial @xmath442-bundle over @xmath250 with the desired monodromy map .
now let us look at how the operation @xmath367 on @xmath39 works . to give a statement in detail
, we divide our discussion into the following three cases . 1 .
if we exactly cut out a 2-independent triangular facet from @xmath37 , then we also need to cut out such a facet from a colored 3-sided prism @xmath457 .
according to the colorings on @xmath63 , the topological type of @xmath338 must be @xmath2 , so we can cut out a @xmath0-invariant @xmath365 from @xmath2 with a certain action @xmath458
. then we glue @xmath459 and @xmath460 along their boundaries , i.e. , @xmath461 2 .
if we exactly cut out a 2-independent square facet @xmath101 from @xmath37 , then we need a colored 4-sided prism @xmath462 to do a coloring change of @xmath101 . in this case , the section in @xmath37 or @xmath462 is a 2-independent square section @xmath171 .
if @xmath171 is 2-colorable ( i.e , @xmath171 corresponds to a disjoint union @xmath323 by lemma [ section ] ) , then by lemma [ 4-prism1 ] @xmath375 is equivariantly homeomorphic to a twisted @xmath316-bundle over @xmath250 , and we can cut out a @xmath0-invariant @xmath368 from @xmath375 .
if @xmath171 is 3-colorable ( i.e , @xmath171 corresponds to a disjoint union @xmath324 by lemma [ section ] ) , then by lemma [ 4-prism1 ] , @xmath463 is equivariantly homeomorphic to a twisted @xmath317-bundle over @xmath250 , and we can cut out a @xmath0-invariant @xmath369 from @xmath375 . combining these arguments , we conclude that if the topological type of @xmath375 is a twisted @xmath316-bundle over @xmath250 , then @xmath464 and if the topological type of @xmath375 is a twisted @xmath317-bundle over @xmath250 , then @xmath465 3 . if we exactly cut out a 2-independent pentagonal facet @xmath101 from @xmath37 , then we need a colored 5-sided prism @xmath466 to change the coloring of @xmath101 . since the section of @xmath117 or @xmath467 is a 2-independent pentagonal section @xmath164 , by lemmas [ section ] and [ 5-prism ]
, @xmath382 is equivariantly homeomorphic to a twisted @xmath442-bundle over @xmath250 , and we can cut out a @xmath0-invariant @xmath468 from @xmath379 . then the operation @xmath367 of @xmath40 and @xmath382 is as follows : @xmath469 in doing the operation @xmath367 on a @xmath40 , we cut out a small facet from @xmath109 and a bottom facet from a colored @xmath209-sided prism @xmath372 , @xmath211 , and then glue them together along their sections .
there are similar procedures for @xmath40 and @xmath470 .
namely , we first remove an open @xmath0-invariant @xmath383-handlebody @xmath471 from @xmath40 and @xmath470 respectively where @xmath383 is a @xmath245 , or a torus , or a klein bottle , or a @xmath443 , and then glue back the remaining part ( i.e. , a @xmath0-invariant @xmath383-handlebody @xmath472 $ ] ) of @xmath470 to @xmath473 along their boundaries .
when @xmath373 , @xmath474 is a @xmath245-bundle over @xmath250 but it is always trivial , so we can glue back the remaining part of @xmath474 to @xmath473 without any twist .
however , when @xmath475 or 5 , since @xmath470 is always a non - trivial bundle over @xmath250 by lemmas [ 4-prism1 ] , [ 4-prism2 ] and [ 5-prism ] , this means that gluing back @xmath472 $ ] actually leads to the appearance of some twist of @xmath472 $ ] , as shown in the following figure : @xmath476 when we do the operations @xmath78 and @xmath367 on @xmath39 , we see that after removing an open @xmath0-invariant desired 3-manifold from @xmath477 or @xmath478 , the remaining part is still a same type of @xmath0-invariant 3-manifold with boundary but admits a different @xmath0-action . of course
, the actions on these two 3-manifolds are compatible with the action on @xmath477 or @xmath478 .
this means that @xmath477 and @xmath478 admit equivariant heegaard splittings ( cf @xcite ) .
stong showed in @xcite that the @xmath479-equivariant unoriented cobordism class of each closed @xmath26-manifold is determined by that of its fixed data .
this gives the following result in the special case .
[ bound ] suppose that a closed manifold @xmath10 admits a @xmath26-action such that its fixed point set is finite
. then @xmath10 bounds equivariantly if and only if the tangent representations at fixed points appear in pairs up to isomorphism .
each @xmath5-dimensional small cover @xmath480 has a finite fixed point set , which just corresponds to the vertex set of @xmath7 .
by gkm theory @xcite , its tangent representations at fixed points exactly correspond to a @xmath481-coloring on the 1-skeleton of @xmath7 .
it is not difficult to check that this @xmath481-coloring on the 1-skeleton of @xmath7 is algebraically dual to the @xmath26-coloring on @xmath7 , as seen in the proof of lemma [ basic small-3 ] .
therefore , we have that the @xmath26-colorings of two vertices @xmath482 in @xmath7 are the same if and only if the corresponding tangent representations at the two fixed points @xmath483 are isomorphic .
moreover , by proposition [ bound ] we conclude that now let us look at how six operations work in @xmath88 . given two classes @xmath484 $ ] and @xmath485 $ ] in @xmath88 ,
when we do the operation @xmath77 on @xmath362 and @xmath363 , we need to cut out two vertices with same coloring from @xmath124 and @xmath145 respectively .
this means that we exactly cancel two fixed points with same tangent representation in @xmath486 , but this does not change @xmath487 up to equivariant cobordism by proposition [ bound ] .
thus we have [ cobordism2 ] let @xmath489 $ ] be a class in @xmath88 .
then @xmath490=[m(p^3,\lambda)]+[m(p^3(3 ) , \tau)]\\ & [ m(p^3,\lambda)\widetilde{\sharp^{eve}}m(p^3_-(3 ) , \tau)]=[m(p^3,\lambda)]+[m(p^3_-(3 ) , \tau)]\\ & [ m(p^3,\lambda)\widetilde{\natural}m(\text{\large $ \oslash$ } , \tau)]=[m(p^3,\lambda)]\\ & [ m(p^3,\lambda)\widetilde{\sharp^\copyright}m(p^3(i ) , \tau)]=[m(p^3,\lambda)]+[m(p^3(i ) , \tau ) ] , i=3,4,5.\end{aligned}\ ] ] lemmas [ cobordism1 ] and [ cobordism2 ] tell us that five operations @xmath491 , @xmath78 , @xmath367 have a nice compatibility with the disjoint union in the sense of equivariant cobordism .
notice that clearly @xmath477 bounds equivariantly by proposition [ bound ] , so @xmath492=0 $ ] in @xmath88 . however , the operation @xmath361 is a little different from other five operations in @xmath88 .
let @xmath484 $ ] and @xmath485 $ ] be two classes in @xmath88 .
when we do the operation @xmath361 on @xmath362 and @xmath363 , it is possible that we just cut out two triangular facets with different colorings from @xmath130 and @xmath145 respectively .
if this happens , then we glue the two parts cut out from @xmath124 and @xmath145 along their sections , so that we can form a 3-sided prism @xmath63 with a natural induced coloring ( denoted by @xmath493 ) such that top and bottom facets are colored differently .
furthermore , this colored 3-sided prism can be recovered into a small cover .
thus , by proposition [ bound ] we have [ cobordism3 ] let @xmath484 $ ] and @xmath485 $ ] be two classes in @xmath88 .
then @xmath494\widetilde{\sharp^\triangle}[m(p_2 ^ 3 , \lambda_2)]\\=&\begin{cases } [ m(p_1 ^ 3 , \lambda_1)]+[m(p_2 ^ 3 , \lambda_2 ) ] & \text{if we cut out two triangular}\\ & \text{facets with same coloring}\\ [ m(p_1 ^ 3 , \lambda_1)]+[m(p_2 ^ 3 , \lambda_2)]+[m(p^3(3),\lambda_1\sharp^\triangle\lambda_2)]&\text{if we cut out two triangular}\\ & \text{facets with different colorings . } \end{cases}\end{aligned}\ ] ] | in this paper we study the ( equivariant ) topological types of a class of 3-dimensional closed manifolds ( i.e. , 3-dimensional small covers ) , each of which admits a locally standard @xmath0-action such that its orbit space is a simple convex 3-polytope .
we introduce six equivariant operations on 3-dimensional small covers .
these six operations are interesting because of their combinatorial natures .
then we show that each 3-dimensional small cover can be obtained from @xmath1 and @xmath2 with certain @xmath0-actions under these six operations . as an application
, we classify all 3-dimensional small covers up to @xmath3-equivariant unoriented cobordism . |
the inflation model is widely accepted as the standard cosmological model which describes the universe from @xmath1 seconds after its creation . however , standard cosmology presents many unresolved problems , including the origin of large scale structure of the universe , the origin of density fluctuations , which initiate galaxy formation , and the mechanism for baryon asymmetry , which leads to the production of more matter than antimatter
. it has been conjectured that many of these problems can be resolved by incorporating topological defects with inflation @xcite .
the most popular defect model exploits the presence of cosmic strings , which are predicted by modern grand unified theories ( guts ) to have formed as the early universe passed through successive phase transitions @xcite . around @xmath2 seconds after the big bang
, the universe passed through a phase transition corresponding to symmetry breaking of the gut force to distinct strong and electroweak forces . at this time
an extensive gut cosmic string network is believed to have formed @xcite .
the subsequent evolution of this network introduces density fluctuations into an otherwise homogeneous universe , which are conjectured to have lead to large scale structure in the observable universe @xcite .
however , cosmic strings are not topologically stable in simple non - abelian models , and therefore are not an inevitable prediction of all gut models .
cosmic strings are only stable in models which involve discrete symmetry breaking @xcite ( e.g. , @xmath3 ) , or when a @xmath4 generator is gauged to form semi - local strings @xcite .
although cosmic strings provide a mechanism for introducing fluctuations , they also pose significant problems for cosmology .
many gut models predict that a cosmic string has an internal structure , which allows it to become superconducting @xcite .
superconducting cosmic string loops can stabilize to vortons @xcite , which would interfere with primordial nucleosynthesis @xcite .
in addition to predicting the existence of cosmic strings , guts also predict large numbers of magnetic monopoles @xcite which are not observed @xcite .
this is often referred to as the monopole problem . however , in models of @xmath5 symmetry breaking , sphalerons form instead of monopoles @xcite .
sphalerons are of interest since they are capable of baryon and lepton number violating processes @xcite . in this paper
we show that it is possible to construct a cosmologically significant defect in @xmath0 models , that behaves locally like a string , but is a topologically stable network texture .
the local stability is a consequence of the interactions between adjacent string defects .
since this new defect is stabilized by the surrounding string network , we introduce the name homilia string to differentiate it from cosmic strings .
sphalerons are identified with the intersection of two homilia strings , and the separation of the strings results in the annihilation of the sphaleron .
furthermore , electroweak homilia string loops can not stabilize as bosonic or fermonic vortons , which circumvents the adverse cosmological problems of stable loops .
homilia strings can be generalized to @xmath6 models of symmetry breaking .
gut scale homilia strings will induce fluctuations in the cosmic microwave background ( cmb ) , which are characterized by greater power at smaller angular scales ( in the anisotropy angular power spectrum ) , compared to the spectrum generated by cosmic strings .
this paper is organized as follows . in sec .
[ sec - lagrangian ] we introduce the lagrangian for @xmath0 symmetry breaking and describe the numerical techniques used to analyze homilia strings .
[ sec - stable ] discusses homilia string stability and sec .
[ sec - vortex ] examines the homilia string vortex solution .
the dynamics of homilia strings are investigated in sec .
[ sec - force ] .
the string dynamics are utilized in sec .
[ sec - cosmo ] to highlight the cosmological ramifications of homilia strings , including sphaleron decay , network evolution and vorton instability .
the main results of the paper are summarized in sec .
[ sec - conclusion ] .
both nonsymmetric and cosmic strings are described by the homotopy group @xmath7 ( i.e. , a non - contractible @xmath8 circle ) . however , nonsymmetric strings @xcite differ from cosmic strings since they do not constitute a line of false vacuum . within the core of a nonsymmetric string the higgs field rotates in isospin space to reconcile an undefined phase . usually string defects in non - abelian models
are not topologically stable because the vacuum symmetry group is larger than the circle @xmath8 that constitutes the string defect .
this implies that the non - simply connected region can be continuously deformed into a simply connected region , removing the topological defect .
however , it is possible to construct a special type of nonsymmetric string ( i.e. , a homilia string ) , that is stable in models of @xmath0 symmetry breaking .
the stability of homilia strings is related to the topology of the string network , not to the topology of an individual string . since a homilia string network can not be deformed continuously to the vacuum , it is topologically stable even in non - abelian models
. we introduce homilia strings in the context of a model of spontaneous @xmath0 symmetry breaking .
the results are readily generalized to describe homilia strings in non - abelian @xmath6 models .
the lagrangian which describes @xmath0 symmetry breaking is written as ( with @xmath9 ) @xmath10 where @xmath11 is an isodoublet , @xmath12 , @xmath13 $ ] , @xmath14 and @xmath15 are the pauli spin matrices ( @xmath16 ) . to describe @xmath0 we also write @xmath17 , where the associated gauge field is @xmath18 .
the equations of motion are obtained by varying the lagrangian ( [ eq - lagrangian ] ) with respect to @xmath19 and @xmath20 , respectively [ eq - eom ] @xmath21 where @xmath22 are the structure constants , which for @xmath5 are components of the fully skew - symmetric tensor @xmath23 . to facilitate the numerical simulations ,
the number of parameters in the model is reduced via the rescaling , @xmath24 , @xmath25 , @xmath26 and @xmath27 .
this results in the equations of motion ( [ eq - eom ] ) having one independent parameter @xmath28 , i.e. , [ eq - eome ] @xmath29 where @xmath30 and @xmath31 $ ] . for convenience
we set the parameter @xmath32 in all numerical calculations , except where stated otherwise . to analyze the stability of homilia strings we first calculate the homilia string vortex solution in the lorentz gauge , using an iterative relaxation algorithm .
this is based on a finite difference approximation , with appropriate two - point boundary conditions .
the gauge invariant properties of the homilia string are exploited in the numerical simulations to identify the formation of homilia strings .
we numerically simulate homilia strings in three dimensions by evolving the fields using an explicit finite difference scheme in the temporal gauge ( @xmath33 ) .
the latter simulations were performed on a @xmath34 lattice , with time step @xmath35 and lattice spacing @xmath36 .
the stability of a homilia string is related to interactions with the surrounding network ; consequently , periodic boundary conditions were imposed on the magnitude of the fields to account for the string network outside the simulation domain .
the gradient of the phase field was constrained to zero at the boundaries , since the string configurations described in this paper exhibited discontinuous phase windings across the boundaries .
the numerical model is modified for simulations of string formation and is this discussed further in sec .
[ sec - force ] .
simulations on different lattice sizes were conducted to ensure that the results are independent of the size of the lattice .
the output was generated by determining which lattice squares contained phase twists ( i.e. , @xmath37 phase rotations ) . the lattice square which contained the twist was marked with a light or dark grey dot , depending on the type of homilia string that was detected .
the dots were then displayed on a three - dimensional plot . to verify that these objects are homilia strings we use the vortex solution and plots of the field magnitude , @xmath38 ; this shows deviations from the vacuum state that are characteristic of homilia strings .
these deviations are associated with a non - zero energy density and hence homilia strings are not gauge dependent objects .
we begin our discussion with an explanation of the homilia string stability mechanism .
the equations of motion and vortex solution are discussed in sec .
[ sec - vortex ] .
the stability of an individual homilia string is a consequence of the boundary conditions imposed on a string in the homilia string network . in models involving @xmath0 symmetry breaking ,
we denote a homilia @xmath39-string by the field configuration @xmath40 where @xmath41 is the topologically non - trivial phase winding for the @xmath39-string , centered at some position @xmath42 ( @xmath43 ) , and
@xmath44 is a constant .
the ansatz ( [ eq - ss ] ) has been utilized previously ( see e.g. , @xcite and @xcite ) .
however , our analysis differs from earlier work since we are interested in the behaviour of eq .
( [ eq - ss ] ) when the homilia string is part of a string network .
the @xmath39-string is prevented from deforming into a simply connected region by the presence of a @xmath45-string .
an homilia @xmath45-string is described in @xmath0 by the field configuration @xmath46 where @xmath47 is a constant independent of @xmath44 and @xmath48 is another non - trivial phase winding for a @xmath45-string located at @xmath49 ( @xmath50 ) .
the @xmath39-and @xmath45-strings are separated by a distance @xmath51 . in fig .
[ fig - vortextwoa ] we have plotted a cross - section of the fields @xmath52 and @xmath53 , for an @xmath39-string ( at @xmath54 ) adjacent to a @xmath45-string ( at @xmath55 ) . in plotting the cross - section of the fields we have replaced the radial coordinate @xmath56 by @xmath57 . to reconcile an undefined phase requires [ eq - bcone ] @xmath58 the total energy of the system is minimized when the deviation of the magnitude , @xmath38 , from @xmath59 assumes its smallest value , whence the system undergoes symmetry breaking at all points .
this requires @xmath60 , and imposes the conditions [ eq - bctwo ] @xmath61 the boundary conditions for an @xmath39-string require @xmath52 to be zero at the center of the @xmath39-string and non - zero at the center of the @xmath45-string .
if we surround an @xmath39-string with @xmath45-strings , then @xmath52 can not be zero everywhere .
since @xmath52 can not be reduced to zero globally , the topological defect can not be deformed to a simply connected vacuum and the @xmath39-string is stable .
a similar argument shows that a @xmath45-string is stable in a network of @xmath39-strings .
it is important to emphasize that @xmath39-and @xmath45-strings can not exist in isolation and are not independent objects .
consider now the situation where an @xmath39-string has a single @xmath45-string adjacent to it . in this case , although @xmath52 is prevented from going to zero on one side of the @xmath39-string , it can go to zero on the other side , and hence the @xmath39-string can unwind .
however , by placing an @xmath39-string in a homilia string network , consisting of both @xmath39-and @xmath45-strings , we have @xmath62 for any given point on the @xmath39-string . here
@xmath63 is the angular dependant shortest distance from the @xmath39-string segment to a @xmath45-string ; @xmath56 and @xmath64 denote the radial co - ordinates of a point in a plane taken through the network that intersects the @xmath39-string in question .
for an arbitrary direction there will be a @xmath45-string located near the @xmath39-string at some distance @xmath63 , whence @xmath65 is non - zero . because @xmath65 is non - zero at a distance @xmath63 from the @xmath39-string for all @xmath64 , the homilia @xmath39-string is a stable defect which can not unwind .
a similar argument is valid for @xmath45-strings , which leads us to conclude that @xmath39-and @xmath45-strings are stable in an @xmath0 homilia string network which is unbounded ( i.e. , an infinite or periodic network ) .
in such a network all the homilia strings will be surrounded by homilia strings of the other type .
this situation is accounted for in numerical simulations by imposing periodic boundary conditions on the magnitude of the fields , which prevents homilia strings at the edge of the lattice from unwinding .
as long as the entire homilia network can not be deformed continuously to the vacuum state individual homilia strings in the network are stable .
if we assume that the ` mexican hat ' potential describing symmetry breaking is symmetric , and the fluctuations initiating the phase transition are random , then after a phase transition there will be phase twists in both components of the @xmath11 isodoublet .
the interaction between these phase twists leads directly to field configurations which are described by homilia strings .
the phase twists are prevented from unwinding because each defect makes it energetically unfavourable for adjacent defects to unwind .
the defects are topologically stable in the network , provided the network consists of the homilia string types permitted by the symmetry breaking group ( e.g. , @xmath39 and @xmath45 strings ) .
the homilia string network constitutes a texture defect which locally behaves like a string .
homilia strings differ from non - topological cosmic strings @xcite , since the magnitude of the field , @xmath38 , is non - zero at the the core of the homilia string ( i.e. , the symmetry is unbroken ) .
therefore homilia strings are not associated with a false vacuum .
homilia strings require the @xmath11 multiplet components to be complex , and hence they only exist in models such as @xmath6 . to describe a homilia string in a model of @xmath0 symmetry breaking we write the @xmath11-field in cylindrical form , i.e. , @xmath66 @xmath67 ( @xmath68 ) are complex fields with magnitude @xmath69 and phase @xmath70 . to describe the homilia string we have chosen @xmath71 ( where @xmath72 indicates the winding number ) and @xmath73 to be constant in the vicinity of the defect .
it is important to emphasize that we have chosen a particular orientation of the higgs field , so that the homilia strings separate out into @xmath74 and @xmath75 components . in this section
we determine the vortex solution for this particular orientation of the higgs field in isospin space ( i.e. , the ` homilia gauge ' ) , for which the two homilia string types separate into separate components of the @xmath11 isodoublet .
we are then able to determine gauge invariant properties of homilia strings , which enables us to identify homilia strings in the three - dimensional network simulations . in the homilia gauge ,
it is apparent that the phase winding of the @xmath39-and @xmath45-strings do not couple through the potential . to determine the vortex solution ,
consider an @xmath39-string described by @xmath76 where we have set the phase of the @xmath75-field to zero , and @xmath77 is the generator for the @xmath39-string in the homilia gauge .
this generator is written as @xmath78.\ ] ] the properties of @xmath77 in eq .
( [ eq - defalpha ] ) can be determined from a taylor expansion , utilizing @xmath79 .
similarly , a @xmath45-string generator can be written as @xmath80.\ ] ] with this particular orientation of the @xmath11-field ( i.e. , the homilia gauge ) , the generators are abelian @xmath81=0.\ ] ] to accommodate gauge fields in the vortex solution we define @xmath82 \nonumber \\ \label{eq - hgauge } & = & \frac{1}{2}\left[\begin{array}{cc } 2 b(r ) & c(r ) e^{i\theta } \\
c(r)e^{-i\theta } & 0 \end{array}\right],\end{aligned}\ ] ] where @xmath83 and @xmath84 are real fields and @xmath85 , @xmath86 and @xmath87 are the pauli spin matrices defined by @xmath88 \\ \sigma^{2 } & = & \left[\begin{array}{cc } 0 & -i/2 \\ i/2 & 0 \end{array}\right ] \\
\sigma^{3 } & = & \left[\begin{array}{cc } 1/2 & 0 \\ 0 & -1/2 \end{array}\right],\end{aligned}\ ] ] with @xmath89 .
the equations governing the homilia string vortex solution are obtained by substituting the ansatz [ eqs .
( [ eq - nonsymaz ] ) and ( [ eq - hgauge ] ) ] into the equations of motion ( [ eq - eome ] ) , to obtain ( in the lorentz gauge ) : [ eq - gaugevortex ] @xmath90 & = & 0 \\ c(r)''-\frac{c(r)'}{r } - e^{2 } \left[2f^{\alpha}(r)f^{\beta}(r)\left(b(r)-1\right)+|\phi|^{2}c(r)\right ] & = & 0 . \end{aligned}\ ] ] where a dash denotes differentiation with respect to @xmath56 and @xmath91 in the case of an @xmath0 homilia @xmath39-string we want the boundary conditions to describe a neighboring @xmath45-string at @xmath92 .
the boundary conditions ( [ eq - bcone ] ) and ( [ eq - bctwo ] ) imply that the magnitudes @xmath52 and @xmath53 are equal at a distance mid - way between the @xmath39-string and the nearest @xmath45-string ( see fig .
[ fig - vortextwoa ] ) .
a cylindrically symmetric solution can be obtained by approximating the distribution of the surrounding @xmath45-strings , at the average distance @xmath93 halfway between the segments of homilia @xmath39-string and the surrounding @xmath45-strings , with @xmath94 . assuming the strings do not overlap , this enables us to write the cylindrically symmetric boundary conditions for the approximate vortex solution of the @xmath39-string as [ eq - sutwobc ] @xmath95_{r=0}=0 \\
f^{\alpha}(r= \zeta/2 ) \approx f^{\beta}(r=\zeta/2).\end{aligned}\ ] ] the boundary conditions ( [ eq - sutwobc ] ) differ from those employed in reference @xcite for conventional nonsymmetric strings , due to the orientation of the @xmath11-field in isospin space ( i.e. , the choice of the homilia gauge ) .
the boundary conditions for @xmath83 and @xmath84 in eqs .
( [ eq - hgauge ] ) , consistent with eqs .
( [ eq - gaugevortex ] ) , are @xmath96
a property of the vortex solution ( [ eq - gaugevortex ] ) is that @xmath38 is not equal to @xmath59 everywhere . to demonstrate this ,
consider a global vortex solution where @xmath97 .
the equations of motion for the case @xmath98 are @xmath99 the solutions to the fields @xmath65 and @xmath100 are written as [ eq - deviation ] @xmath101 where @xmath102 , @xmath103 , @xmath104 and @xmath51 are constants .
the boundary conditions ( [ eq - sutwobc ] ) imply that @xmath105 and @xmath106 , and hence the solutions ( [ eq - deviation ] ) can not satisfy @xmath98 everywhere .
we conclude that the presence of a homilia string forces the @xmath38-field to depart from the vacuum state . for homilia strings with gauge fields , the numerical calculations indicate that the result is still valid at the center of the string , where @xmath83 and @xmath84 are small .
this deviation is what gives homilia strings their independent energy density , and demonstrates that they are not equivalent to the vacuum state .
consequently , homilia strings have a tension which causes string loops to collapse .
collapsing loops , together with intercommuting , provide homilia string networks with an energy loss mechanism in the same manner as @xmath4 cosmic string networks @xcite .
this prevents the homilia string network from dominating the energy density of the universe . in fig .
[ fig - gsvortex ] we have plotted an @xmath0 homilia string vortex solution for @xmath107 . the vortex solution was calculated numerically from the equations of motion ( [ eq - gaugevortex ] ) , subject to the boundary conditions ( [ eq - sutwobc ] ) .
the deviation of @xmath38 from @xmath59 at the center of the string is evident in fig .
[ fig - deviation ] , which shows that a deviation is present when gauge fields are included .
this deviation results in a non - zero energy density for the homilia string .
although eqs .
( [ eq - deviation ] ) show that there must be a deviation of the @xmath38-field from the vacuum state , they do not indicate the size of the deviation .
in the case of global @xmath5 symmetry breaking , the size of the deviation depends on the distance between two homilia strings .
as the average distance between global homilia strings increases , the energy per unit length of each global homilia string decreases and quickly becomes negligible .
however , in the case of a homilia string interacting with a gauge field ( e.g. , @xmath0 ) , the gauge field screens the defect in the same way that cosmic string interactions are screened for large @xmath56 in local @xmath4 symmetry breaking models @xcite .
this means that for local homilia strings the deviation of @xmath38 from the vacuum state does not depend on the average separation distance , but rather on @xmath28 . increasing the relative coupling strength results in the gauge field approaching its vacuum solution more rapidly with distance from the string core , and increases the deviation of the @xmath38-field from the vacuum at @xmath108 . for @xmath32
the maximum deviation of @xmath38 from @xmath59 is about @xmath109 , but for @xmath107 the deviation is @xmath110 .
estimates of the higgs boson mass suggest that @xmath111 gev @xcite , and hence this places an upper bound on @xmath28 : @xmath112 where @xmath113 denotes the mass of the intermediate vector boson .
if we assume that the dominate contribution to the energy per unit length is from the potential term , we can establish an upper bound on the electroweak ( @xmath114 gev ) homilia string mass per unit length ( from the vortex solution as ) @xmath115 this is several order of magnitude smaller than an ` electroweak cosmic string ' ( i.e. , false vacuum string ) , which has a mass per unit length of @xmath116 g / m .
the stability of homilia strings in three dimensions was investigated for two homilia strings orientated perpendicular to each other . in fig .
[ fig - gsstable ] , we have plotted two ( initially stationary ) homilia strings in an @xmath0 model .
the three - dimensional simulations were conducted in the temporal gauge ( @xmath33 ) .
homilia strings were initially wound into the fields using the vortex solution ( modified for the temporal gauge ) , with multiple homilia strings , separated by a reasonable distance , superposed in the homilia gauge using the ansatz @xmath117 where @xmath118 and @xmath119 denote the fields of the individual homilia strings and @xmath120 is the number of superposed strings .
the factor @xmath121 is used to normalize the superposed fields . to account for the surrounding homilia string network in fig .
[ fig - gsstable ] , periodic boundary conditions are imposed . the @xmath39-string is initially orientated parallel to the @xmath122-axis and the @xmath45-string is orientated parallel to the @xmath123-axis . after evolving the fields ,
the @xmath39-and @xmath45-strings were found to be stable , suggesting that homilia strings are stable for arbitrary orientations in three dimensions .
the stability of homilia strings is examined further in fig .
[ fig - gsform ] , where we plot the results of a simulation of one hundred nucleating bubbles of random phase and position , originating from a @xmath0 first order phase transition . for this simulation , we have set @xmath107 , @xmath124 and changed the size of the simulation to a @xmath125 cubic lattice .
the boundary conditions were modified to be periodic in the magnitude and phase of @xmath11 . as a consequence of the violent behaviour of @xmath38 , arising from bubble nucleation and string crossing ( see sec . [ sec - sphaleron ] ) , we have added a damping term , @xmath126 , to the equations of motion ( [ eq - eome ] ) , where @xmath127 is the damping factor . for the simulation depicted in fig .
[ fig - gsform ] , the homilia strings were evolved relativistically ( @xmath128 ) for @xmath129 . for @xmath130
the damping factor was increased smoothly to @xmath131 , which made deviations in @xmath38 from the vacuum , corresponding to homilia strings ( fig .
[ fig - deviation ] ) , clearly visible .
ideally we would choose the damping term to be as small as possible ; however , due to the finite size of the simulation , if the damping term is too small the string density is reduced to zero ( via intercommuting and loop production ) , before the field settles sufficiently for deviations in @xmath38 to be clearly distinguished .
moreover , in sec .
[ sec - sphaleron ] we show that the intersection of homilia @xmath39-and @xmath45-strings are associated with the formation and annihilation of sphalerons ( i.e. , points of false vacuum ) .
hence , the damping factor must remain non - zero for the entire duration of the simulation as the formation and annihilation of sphalerons introduces fluctuations in @xmath38 .
the resultant behaviour of the fields after the @xmath0 phase transition is plotted in fig .
[ fig - gsform ] , which shows the formation of several homilia strings of type @xmath39 and @xmath45 .
homilia strings are stable even in a complicated network . in fig .
[ fig - gsform ] , we can not anticipate the local orientation of @xmath11 in isospin space .
consequently , after formation the @xmath39-and @xmath45-strings will not separate into distinct components of the @xmath11 isodoublet , and hence we can not trivially distinguish homilia strings of different types . therefore , both string types have been shaded black in fig .
[ fig - gsform ] .
homilia strings are still observed to form when the damping term is removed , although this makes the interpretation of the behaviour of @xmath38 ambiguous due to its violent behavior . in fig .
[ fig - hsdev ] we have plotted the magnitude of the @xmath11-field for a constant slice in fig . [ fig - gsform ] ( i.e. , for @xmath132 ) .
deviations of @xmath38 from the vacuum of @xmath133 , corresponding to homilia strings are clearly observed .
consequently , homilia strings are associated with gauge invariant properties ( i.e. , a deviation in @xmath38 which results in a non - zero string mass per unit length ) , and are not gauge dependent objects .
the discrepancy between the observed deviations of @xmath133 and the predicted deviation of @xmath134 from the vortex solution can be attributed to the non - uniform distribution of the strings .
this results in the cylindrically symmetric vortex solution ( [ eq - gaugevortex ] ) being a poorer approximation to the defect field behaviour when the average separation is small .
the deviations are expected to become better approximated by the vortex solution as the average separation distance increases with time ( see sec .
[ sec - network ] ) . to analyze a realistic homilia string network , with large average separation distances ,
requires a significant increase of the lattice size in the numerical simulation . nevertheless , fig .
[ fig - hsdev ] shows that homilia strings are associated with gauge invariant properties , in particular a deviation in @xmath38 which contributes to a non - zero string energy per unit length . in fig .
[ fig - gsstable ] it is observed that two homilia strings ` bend ' under the influence of a repulsive ( interaction ) force between the strings .
homilia strings which are of different types ( i.e. , @xmath39 and @xmath45 strings ) will always experience a repulsive force .
this is because the force is related to the stability of homilia strings , originating from the formation of a sphaleron when the two strings intersect ( sec .
[ sec - sphaleron ] ) .
the angular interaction force on a segment of homilia @xmath39-string due to a segment of @xmath45-string is calculated from @xmath135 where @xmath136 is the energy density of the string along the line @xmath137 joining the two string segments ; @xmath51 represents the separation between the segments . in the case of an homilia @xmath39-string
adjacent to a @xmath45-string ( fig .
[ fig - vortextwoa ] ) , the field @xmath65 is zero at the center of the @xmath39-string and non - zero at the center of the @xmath45-string . when the @xmath39-and @xmath45-strings are in close proximity ( for small @xmath51 ) , the boundary conditions can not be satisfied without increasing the energy of the system . equation ( [ eq - cforce ] )
implies that a increase in the energy corresponds to a repulsive force . in the case of cosmic strings , the interaction between strings can be attractive ( for a certain choice of parameters ) , due to the gauge field overscreening the force .
these strings are referred to as type i @xcite .
however , the overscreening depends on the non - trivial phase winding of the two cosmic strings being described by the same generator ( i.e. , the phase fields are the same ) .
this is not the case for homilia strings of different types , which can be seen for strings in the homilia gauge ( sec .
[ sec - vortex ] ) ; here the @xmath39-and @xmath45-strings are wound in different @xmath67-fields ( i.e. , @xmath74 and @xmath75 for @xmath39-and @xmath45-strings , respectively ) .
consequently , the force between homilia strings of different types is always repulsive . equation ( [ eq - cforce ] ) gives the angular force on a @xmath39-string segment due to an adjacent segment of @xmath45-string , i.e. , @xmath138 for @xmath139 , the two string segments are superposed to form a point of false vacuum , @xmath140 ( see sec .
[ sec - sphaleron ] ) .
this means that the force will be a maximum ( as @xmath141 ) at the maximum deviation of @xmath38 from the vacuum .
moreover , @xmath142 is finite , where @xmath143 and consequently the repulsive force is finite . given sufficient velocity , homilia strings of different types can collide and pass through each other .
the different types of homilia strings do not intercommute because the mexican hat potential for an @xmath0 model has the form @xmath144^{2}.\ ] ] only the magnitudes , @xmath65 and @xmath100 , couple through the potential , not the phase fields @xmath145 associated with the two perpendicular homilia string generators . in a previous paper we argued that intercommuting is a consequence of phase dynamics @xcite . since the phase fields of homilia strings of different types do not interact directly ( see sec .
[ sec - vortex ] ) , the homilia strings do not intercommute and instead pass through each other . in the case of a homilia string network ,
homilia strings of different types initially become entangled , which results in a rapid increase in the string tension .
the increased tension compensates for the repulsive force and the homilia strings pull through each other .
therefore , homilia string networks will only become entangled for brief intervals of time .
this conjecture is supported by the network formation simulation ( fig .
[ fig - gsform ] ) , where @xmath39-and @xmath45-strings are observed to pass through each other in a collision . in order to produce a stable @xmath0 homilia string
, we require a string of one type to be surrounded by the other string type ( e.g. , an @xmath39-string surrounded by @xmath45-strings ) . however , after formation the homilia string network will be quite complicated and it is possible that a homilia string will be adjacent to a homilia string of the same type .
homilia strings of the same type interact rapidly , as they are unstable and locally unwind ( i.e. , @xmath146 between two strings ) .
such a scenario is depicted in fig .
[ fig - dynamics ] , where two homilia @xmath39-strings , initially orientated perpendicular to each other , are observed to be unstable .
the extra homilia @xmath45-string serves to prevent the entire system from globally deforming to the vacuum . the two @xmath39-strings display complicated dynamics , with string segments moving towards each other and colliding . as the phase twists of the two @xmath39-strings are wound in the same @xmath147-phase ( associated with the homilia @xmath39-string generator ) , the two @xmath39-strings intercommute and move apart due to string tension .
therefore , homilia string networks can form loops , which subsequently collapse due to string tension .
this allows homilia strings to reduce the energy of the string network , preventing homilia strings from dominating the universe .
this mechanism is identical to that for @xmath4 cosmic string networks @xcite .
sphalerons are point defects of false vacuum that form in models of @xmath5 symmetry breaking @xcite
. however , a topologically stable string network in @xmath0 produces sphalerons in the context of homilia strings .
consider the intersection between a homilia @xmath39-and @xmath45-string , where the @xmath39-string is parallel to the @xmath122-axis and the @xmath45-string is parallel to the @xmath123-axis and the point of intersection is at @xmath148 .
we write the @xmath11-field for @xmath0 in the form @xmath149 where @xmath150 , @xmath151 and @xmath152 ( @xmath68 ) . in the homilia gauge ( sec .
[ sec - vortex ] ) , the intersection of the two homilia strings is determined by the phase winding [ eq - monoconst ] @xmath153 since @xmath154 along the @xmath39-string ( @xmath155 ) and @xmath156 along the @xmath45-string ( @xmath157 ) , at the point of intersection ( @xmath148 ) we have @xmath158 .
this is a point of false vacuum and can be interpreted as a sphaleron . since a sphaleron represents the intersection of an @xmath39-and @xmath45-string , it is not an independent object . in sec .
[ sec - force ] we argued that homilia strings of different types exhibit a repulsive force for any choice of parameters .
this is because the energy associated with the sphaleron configuration is much greater than that of separate homilia strings , which results in a repulsive force .
the repulsive force is finite as the energy of the sphaleron is finite .
consequently , the sphaleron configuration of two intersecting homilia strings ( [ eq - monoconst ] ) is inherently unstable . given a perturbation ( e.g. , a kink propagating along one of the strings ) ,
the two homilia strings will separate , resulting in sphaleron decay ( see fig .
[ fig - sphaleron ] ) .
perturbations in the string network can be generated by a number of mechanisms , including kinks arising from intercommuting and the collapse of homilia string loops .
collapsing loops pull the two homilia strings apart annihilating the sphaleron .
sphalerons which form after a @xmath0 phase transition can be viewed as the points of intersection between homilia @xmath39-and @xmath45-strings . as the homilia string network evolves ,
the intersecting homilia strings separate , annihilating the points of false vacuum and the sphalerons decay . to prevent the network from becoming entangled
, homilia strings of different types will sometimes pass through each other ( see sec .
[ sec - force ] ) .
such events correspond to the rapid formation and annihilation of sphalerons .
however , since sphalerons are related to the intersections of strings in the homilia string network , the sphaleron density will decrease as the homilia string network density decreases ( see sec .
[ sec - network ] ) . in the electroweak model ,
the @xmath0 lie generators are fixed in isospin space relative to the higgs vacuum , i.e. , @xmath159 regardless of the actual state , we can rotate @xmath11 in isospin space to @xmath160 , using a local gauge transformation @xmath161 , i.e. , @xmath162 @xmath161 is expressed in its most general form by @xmath163 where @xmath164 is a constant matrix that describes a global gauge transformation , @xmath165 is a real vector and @xmath166 is a vector corresponding to the pauli spin matrices . consider the form of the homilia @xmath39-string ( [ eq - defalpha ] ) at large distances from the string core . in terms of the boundary conditions ( [ eq - sutwobc ] ) , when @xmath167 the @xmath39-string is given by @xmath168 to describe the electroweak vacuum state at large distances , we introduce a gauge transformation @xmath169 for the far field of an homilia @xmath39-string to be consistent with the electroweak vacuum state , we must perform a gauge transformation of the form @xmath170 so that @xmath171 note that @xmath172 is not defined at @xmath108 , where @xmath64 is undefined .
although we can locally gauge away the phase variation in @xmath11 , we can not gauge away variations in the magnitudes of the @xmath11 multiplet [ i.e , @xmath65 and @xmath173(r ) ] . in the limit as @xmath174 ,
the boundary conditions require @xmath175 and @xmath176 , where @xmath177 is the maximum deviation of @xmath38 from the vacuum .
therefore , as @xmath174 , in the vicinity of the @xmath39-string we find @xmath178 by choosing a gauge so the @xmath11-field is consistent with the electroweak model far from the homilia string , @xmath11 must rotate in isospin space away from @xmath160 as @xmath174 .
a similar analysis applies to @xmath45-strings and consequently homilia @xmath39-and @xmath45-strings can not be gauged away .
this rotation has implications for the superconductivity of homilia strings discussed in sec .
[ sec - sc ] .
cosmic string defects can have an internal structure and become superconducting @xcite .
superconducting cosmic strings are classified according to whether they have bosonic , fermonic or non - abelian gauge supercurrents .
however , superconducting cosmic string loops have adverse cosmological consequences since they stabilize as vortons @xcite , which interfere with primordial nucleosynthesis @xcite .
homilia strings avoid these problems , since the behaviour of the supercurrent inside a homilia string is different to that of a cosmic string . to show this we examine superconductivity in homilia strings .
bosonic supercurrents couple to the false vacuum through the potential @xcite @xmath179 where the complex scalar field @xmath180 represents the bosonic supercurrent . to form a stable string
imposes the condition @xmath181 to prevent the supercurrent from becoming non - zero outside the string core imposes an additional constraint @xmath182 at the center of the homilia @xmath183 , therefore a non - zero current can only exist at the homilia string core for @xmath184 combining eqs .
( [ eq - conone ] ) and ( [ eq - contwo ] ) yields @xmath185 depending on the choice of @xmath28 , eq .
( [ eq - boson ] ) indicates that homilia strings will only support bosonic supercurrents for a narrow range of parameters , and in general homilia strings will not support bosonic superconductivity .
furthermore , the vorton current is unstable and will drift off the string core unless @xcite @xmath186 where @xmath187 is a constant that represents the current phase winding number and @xmath188 is the radius of the vorton . from eq .
( [ eq - boson ] ) and eq .
( [ eq - drift ] ) we find that @xmath189 the quantity @xmath190 can be determined @xcite , to give @xmath191 where @xmath192 is the unperturbed string width . combining eq .
( [ eq - ratio ] ) with eq .
( [ eq - compare ] ) results in @xmath193 which clearly contradicts eq .
( [ eq - nocurrent ] ) when @xmath194 . since there is an empirical bound on @xmath28 , in the electroweak model , given by @xmath195 ( sec .
[ sec - vortex ] ) we write @xmath196 and stable bosonic superconducting electroweak homilia vortons are ruled out . to describe a homilia vorton with a fermonic current
, we write the field for a homilia string ( e.g. , an @xmath39-string ) as @xmath197 where the phase of the @xmath75-field has been set to zero .
fermionic currents are coupled to the @xmath11-field through the yukawa interaction .
if we include neutrinos and neglect gauge fields , the interaction lagrangian is @xmath198 where @xmath199 and @xmath200 are components of the left - handed isodoublet , @xmath201 is the right - handed isosinglet and @xmath202 denotes the coupling constant . to incorporate the homilia string into eq .
( [ eq - fermionic ] ) , we must rotate the @xmath11-field so that @xmath203 .
since we have neglected gauge fields the rotation is achieved using the global gauge transformation @xmath164 in eq .
( [ eq - ewtrans ] ) .
this transform results in @xmath204 after performing the gauge transformation , we can substitute the forms of @xmath205 and @xmath206 into the equations of motion for the lepton fields .
the equations of motion are @xmath207 because the @xmath11-field rotates in isospin space inside the homilia string core ( i.e. , @xmath205 and @xmath206 are non - zero ) , the fields @xmath200 , @xmath201 and @xmath199 all become coupled .
we also note that in general @xmath208 . from eq .
( [ eq - fsc ] ) , we derive the relationship @xmath209 multiplying both sides of eq .
( [ eq - psirel ] ) by @xmath210 we obtain @xmath211 equation ( [ eq - psirelx ] ) has the solution [ eq - psisoln ] @xmath212 where @xmath213 is a constant , @xmath214 describes the spinor properties of @xmath200 and @xmath199 and @xmath215 is a non - zero function
. however , if we multiply eq .
( [ eq - psirel ] ) by @xmath216 we obtain @xmath217 substituting eqs .
( [ eq - psisoln ] ) into eq .
( [ eq - psirely ] ) results in the condition @xmath218 to solve eq .
( [ eq - psirely ] ) we utilize the fact that @xmath205 and @xmath206 are non - zero functions , determined by the vortex solution . since @xmath208 , the solution to eq .
( [ eq - condition ] ) requires ( for @xmath219 ) @xmath220 which can not be satisfied in general .
therefore , eq . ( [ eq - condition ] ) requires @xmath221 and this results in the solutions for the lepton fields having zero values , i.e. , @xmath222 .
homilia strings can not support fermonic supercurrents which is a direct consequence of the rotation of the @xmath11-field in isospin space , which leads to the coupling of the fermionic current to neutrino fields .
although homilia strings are expected to support non - abelian gauge supercurrents @xcite , such supercurrents are unlikely to support vorton states .
this is because vorton stability is associated with conserved quantum numbers related to the noether charge and to current phase winding in the loop @xcite .
such quantum numbers do not appear to exist for non - abelian gauge supercurrents @xcite , and hence non - abelian gauge superconducting vortons are unlikely .
since homilia string loops do not support fermionic superconductivity or bosonic vortons , they do not settle into vorton states .
hence homilia strings will avoid the adverse cosmological consequences of stable loops .
if symmetry breaking of an @xmath6 model occurs at the gut phase transition , gut scale homilia strings will also form .
in such models there are @xmath223 types of homilia strings , which after performing an appropriate gauge transformation are wound in the @xmath223 complex components of the @xmath11 multiplet . in a network of homilia strings , consisting of all @xmath223 types of homilia strings , the network texture is topologically stable . due to the number of possible gauge fields , the vortex solution of such a configuration depends on the order of the symmetry breaking group .
simulations of global homilia strings in @xmath224 symmetry breaking indicate that they are stable against collapse into @xmath5 global homilia strings .
this suggests that gut homilia strings would be stable . for symmetry groups of the form @xmath6
, the formation of sphalerons requires the intersection of all @xmath223 types of homilia strings at a point .
such an intersection is unlikely for @xmath225 and hence we do not expect gut sphalerons to form .
gut homilia strings retain the same useful cosmological properties as cosmic strings gut models which do not predict the formation of stable cosmic string defects can nevertheless accommodate large scale structure and density fluctuations induced by gut homilia strings .
since homilia strings have a mass per unit length , cylindrical symmetry and are lorentz invariant along the string length , they will also exhibit a conical spacetime similar to cosmic strings @xcite .
given a gut phase transition at an appropriate scale , it is therefore possible for homilia strings to produce large scale structure through ` wakes ' @xcite and collapsing string loops @xcite .
the evolution of a homilia string network can be described in a manner identical to @xmath4 cosmic string networks @xcite .
loops arise from intercommuting homilia strings , with subsequent collapse and annihilation providing an energy loss mechanism for the network @xcite . in this way
, an homilia string network is prevented from dominating the energy density of the universe . since homilia strings only intercommute with strings of the same type ( sec .
[ sec - force ] ) , we can model an homilia string network consisting of different string types as the superposition of two ( or more ) independent networks . we write @xmath226 where @xmath227 is the total energy density of infinitely long homilia strings , @xmath223 is the order of the symmetry breaking group , which leads to @xmath223 types of homilia strings , and @xmath228 is the energy density of the strings for an independent network of strings of the same type .
the energy density of infinite strings in a standard cosmic string network would also be described by @xmath228 , which for large @xmath229 approaches the scaling solution @xmath230 .
comparing the characteristic length scale of homilia strings , @xmath231 , in the network to that of cosmic strings we find @xmath232 where @xmath233 is the energy per unit length of a homilia string and @xmath234 is the characteristic length scale of a cosmic string network . because homilia strings are inefficient at intercommuting and loop formation ( i.e. , they only intercommute with strings of the same type ) , the characteristic length scale of strings in an homilia string network is smaller than that for strings in a cosmic string network , at a given time @xmath229 .
equation ( [ eq - lensol ] ) indicates that the number density of a network of homilia strings will be greater than the corresponding value for a cosmic string network .
the evolution of an homilia string network leaves characteristic fluctuations in the cosmic microwave background ( cmb ) , providing an experimental test for homilia strings .
since homilia strings possess a conical metric , gut homilia strings will be capable of inducing non - gaussian fluctuations in the cmb in the same manner as cosmic strings @xcite .
the distance between homilia strings , @xmath231 , is proportional to @xmath235 [ see eq .
( [ eq - lensol ] ) ] , hence cmb fluctuations induced by homilia strings will have a smaller angular size than for cosmic strings . based on estimates of cosmic string fluctuations @xcite , the angular size of the fluctuations induced by homilia strings , at the time of last scattering , is given approximately by @xmath236 where @xmath237 is the redshift at the surface of last scattering ( @xmath238 ) .
hence we would expect a gut scale homilia string network to provide much greater power in the angular power spectrum of the cmb anisotropy ( at small angular scales ) , than a network of cosmic strings .
measurements of the anisotropy angular power spectrum may provide a means of differentiating fluctuations due to homilia strings from those induced by cosmic strings .
however , to quantify the magnitude of these fluctuations will require a detailed simulation of an homilia string network .
in this paper we have introduced a new type of topological string defect , called a homilia string , which arises in @xmath6 models of symmetry breaking .
homilia strings are classically stable in @xmath0 , consequently , they are a prediction of the standard electroweak unification model .
individual homilia strings are stabilized via their boundary conditions with adjacent defects .
the topological stability of these string defects is a consequence of the surrounding homilia string network , which can be considered to be a texture .
although some aspects of homilia string dynamics are similar to @xmath4 cosmic strings , homilia strings have additional properties which may have important consequences for cosmology . sphalerons in @xmath0 are identified with the intersection of two homilia strings . due to repulsive forces , homilia strings separate and induce sphaleron decay .
electroweak homilia strings loops do not support bosonic or fermionic vorton states and hence avoid the adverse consequences associated with vortons .
homilia strings retain the advantages of gut cosmic strings for inducing large scale structure in the observable universe and non - gaussian fluctuations in the cosmic microwave background . the anisotropy angular power spectrum of homilia strings is expected to exhibit greater power , at small angular scales , than the spectrum originating from cosmic strings .
future work will be directed at examining the detailed cosmological ramifications of homilia strings , and their potential for experimental verification .
investigating homilia strings in more complex guts , which support cosmic strings ( e.g. , @xmath239 ) , would also allow us to study the interactions between homilia strings and cosmic strings . to understand in detail the differences between homilia strings and cosmic strings will require detailed simulations of large scale homilia string networks .
a. vilenkin and e. shellard , _ cosmic strings and other topological defects _ , ( cambridge university press , cambridge , england , 1994 ) t. kibble ,
_ j. phys .
a _ * 9 * , 1387 ( 1976 ) .
a. vilenkin , _ phys .
reports _ * 121 * , 263 ( 1985 ) .
a. stebbins , s. veeraraghavan , r. brandenberger , j. silk and n. turok , _ ap . j. _ * 322 * , 1 ( 1987 ) .
t. vachaspati and a. vilenkin , _ phys .
d _ * 67 * , 1057 ( 1991 ) .
a. davis and s. davis , _ phys .
d _ * 55 * , 1879 ( 1997 ) .
t. vachaspati and a. achucarro , _ phys .
d _ * 44 * , 3067 ( 1991 ) .
e. witten , _ nucl .
phys . _ * b249 * , 557 ( 1985 ) .
a. everett , _ phys .
lett . _ * 61 * , 1807 ( 1988 ) .
r. davis and e. shellard , _ phys .
b _ * 209 * , 485 ( 1988 ) .
r. brandenberger , b. carter , a. davis and m. trodden , _ phys .
* 54 * , 6059 ( 1996 ) .
g. t hooft , _ nucl .
phys . _ * b79 * , 276 ( 1974 ) . j. preskill , _ phys . rev
_ * 43 * , 1365 ( 1979 ) .
f. klinkhamer and m. manton , _ phys .
d _ * 30 * , 2212 ( 1984 ) .
e. leader and e. predazzi , _ an introduction to gauge theories and modern particle physics _ , ( cambridge university press , cambridge , england , 1996 ) m. axenides and l. perivolaropoulos , _ phys .
d _ * 56 * , 1973 ( 1997 ) .
m. hindmarsh , _ phys .
lett . _ * 68 * , 1263 ( 1992 ) .
p. saffin and e. copeland , _ phys .
d _ * 56 * , 1215 ( 1997 ) .
e. shellard , _ nucl .
* b283 * , 624 ( 1987 ) .
e. myers , c. rebbi and r. strilka , _ phys .
d _ * 45 * , 1355 ( 1992 ) .
m. thatcher and m. morgan , _ class .
* 14 * , 3161 ( 1997 ) .
r. davis _ phys .
d _ * 38 * , 3722 ( 1988 ) .
t. kibble , g. lozano and a. yates , _ phys .
d _ * 56 * , 1204 ( 1997 ) .
w. hiscock , _ phys .
d _ * 31 * , 3288 ( 1985 ) .
l. perivolaropoulos , r. brandenberger and a. stebbins , _ phys .
d _ * 41 * , 1764 ( 1990 ) .
r. scherrer , a. melott and e. bertschinger , _ phys .
lett . _ * 62 * , 379 ( 1989 ) .
n. kaiser and a. stebbins , _ nature _ * 310 * , 391 ( 1984 ) .
f. bouchet , d. bennett and a. stebbins , _ nature _ * 335 * , 410 ( 1988 ) . | in this paper we describe a new type of topological defect , called a homilia string , which is stabilized via interactions with the string network . using analytical and numerical techniques
, we investigate the stability and dynamics of homilia strings , and show that they can form stable electroweak strings . in @xmath0 models of symmetry breaking the intersection of two homilia strings is identified with a sphaleron . due to repulsive forces , the homilia strings separate , resulting in sphaleron annihilation .
it is shown that electroweak homilia string loops can not stabilize as vortons , which circumvents the adverse cosmological problems associated with stable loops .
the consequences for gut scale homilia strings are also discussed . |
the orbital degrees of freedom play important role in the condensed matter physics to understand the various phenomena observed in the strongly correlated systems from last few decades@xcite .
transition metals having partially filled degenerate @xmath5 orbitals , where the certain kind of orbital being occupied at specific site is expected to be directed by the another kind of orbital being occupied at another site .
hence such type of preferred occupation of orbitals in a system can lead to the various kind of orbital ordering ( oo ) .
the oo is strongly linked with the spin ordering ( so ) in many compounds . in order to understand
the various so observed in la@xmath6ca@xmath7mno@xmath8 compounds goodenough has proposed different kinds of oo@xcite .
the famous goodenough - kanamori - anderson rules provide the scheme which help in predicting the possible oo for a given so observed in a compound@xcite . in transition metal oxides coulomb repulsion _
u _ are found to play an important role in driving the oo .
however it is still unclear whether _ u _ only enhances the effect of lattice distortions or really drives the oo in the transition metal oxides via super - exchange mechanism@xcite .
oo is also responsible for the structural transition as proposed in the spinel vanadates@xcite .
according to tchernyshov the structural transition arises in the spinels from cubic to tetragonal due to the interplay of the jahn - teller effect and relativistic spin - orbit interaction@xcite .
et al . _ suggested a model that predicts the structural ground state due to the anti - jahn teller effect caused by the broadening of the @xmath9 , @xmath10 bands and leads to the orbitally driven peierls state@xcite .
however tsunetsugu _ et al .
_ showed that the coulomb and exchange interaction between the magnetic ions and the coupling to the jahn - teller lattice distortion are responsible for the structural transition in the spinels@xcite .
oo also affects the transport behavior of a compound because of the localization of the electrons in the orbitals .
the occupancy of orbitals and their orientation in the space leads the anisotropic behavior in the system@xcite . before 1990 s _ ab initio _
density functional theory was not a popular approach for calculating the oo in a strongly correlated system . for most of the materials
the oo were calculated by using the model calculations@xcite .
however with the implementation of lda+_u _ scheme in the _ ab initio _ density functional theory based codes the study of oo in many materials became popular . in this approach
the density matrices ( dm ) of the correlated orbitals are the natural output of the lda+_u _ calculations .
these dm are used for knowing the orbital occupancies of orbitals .
most of the codes calculates the dm in the spherical harmonic representation .
normally the oo is shown in cubic harmonic representation . moreover , knowing the oo in cubic harmonics looking at the dm in the spherical harmonics is not a straightforward job .
this becomes more complex if the transition metal is not in the regular octahedral symmetry . in this work we use the dm of vanadium atoms to study the oo pattern of the spinel vanadates , which has been a contentious issue for a long time@xcite .
spinel vanadates with general formula av@xmath0o@xmath1 ( a @xmath2 zn , cd and mg ) possessing cubic structure at room temperature and forming pyrochlore lattice at the v sites , which is responsible for the geometrical frustration@xcite .
at low temperature these compounds show structural transition from cubic to tetragonal@xcite .
the driving force for the structural transition is proposed to be the oo@xcite .
different studies show different oo in these compounds .
for example , based on model calculations , motome _ et al .
_ have predicted the anti - ferro oo , where @xmath3 and @xmath4 orbitals are mainly occupied alternatively along c - axis together with the occupancy of @xmath11 orbital at every v sites@xcite .
this oo is inconsistent with the spatial symmetry _
i_4@xmath12/_amd _ of tetragonal phase as it breaks the mirror reflections in the planes ( 110 ) and ( 1@xmath130 ) and diamond glides _
d _ in the planes ( 100 ) and ( 010)@xcite . however , in the presence of strong relativistic spin - orbit coupling ( soc ) and inclusion of jahn - teller effect as a perturbation the model calculations of tchernychyov have predicted a complex uniform ferro - orbital ordering of type @xmath3@xmath14i@xmath4 at each v sites@xcite .
this oo is consistent with the spatial symmetry _
i_4@xmath12/_amd _ as it preserves both mirror and glide symmetries .
proposed a theoretical model in which the spin - orbital super - exchange interaction and soc have been treated on equal footing .
they have shown the existence of complex orbital ordered ground state similar to that of tchernychyov for intermediate spin - orbit coupling strength with the symmetry _
proposed a mechanism considering the superstructure formed in spinel containing transition metal ions .
they have shown that close to the itinerant state the orbital - peierls ordering of type @xmath4-@xmath4-@xmath3-@xmath3 along the tetragonally compressed c - axis have been observed@xcite .
based on 3-dimensional electron spin - density plots obtained from lsda+_u _ calculations on znv@xmath15o@xmath16 compound , maitra _
have shown the existence of @xmath3+@xmath4 and @xmath3-@xmath4 oo in the absence of soc . in the presence of soc they found the mixture of @xmath3+i@xmath4 and @xmath3-i@xmath4 orbitals@xcite . in order to study the oo of av@xmath0o@xmath1 ( a @xmath2 zn , cd and mg ) compounds
, we use the dm ( in the spherical harmonics representation ) of vanadium atoms obtained from gga+_u _ calculations .
our studies in the absence of soc suggest the anti - ferro oo in the global ( local octahedral ) coordinate system for all the compounds where @xmath3 and @xmath4 ( @xmath3+@xmath4 and @xmath3-@xmath4 ) orbitals are mainly occupied at the neighboring v sites .
the dm of vanadium atoms in znv@xmath0o@xmath1 , mgv@xmath0o@xmath1 and cdv@xmath0o@xmath1 compounds are calculated by using the _ state - of - the - art _ full - potential linearized augmented plane wave ( fp- lapw ) method @xcite .
the lattice parameters and atomic positions used here for every compounds are taken from the literature@xcite .
pbesol exchange correlation functional is used in the calculation @xcite .
the effect of on - site coulomb interaction among v 3d electrons is considered within gga+_u _ formulation of the density functional theory @xcite .
fully localized double counting scheme has been considered in the calculations@xcite . in this method _ u _ and _ j _ are used as parameters .
we have varied _ u _ from 3 ev to 5 ev at fixed _ j _
= 0.5 ev and obtained the dm of v atoms in these compounds which are shown in the tables 19 .
one can see a slight change in the values of the dm when _ u _ increases from 3 to 4 ev .
it remains almost the same for _ u _ = 4 - 5 ev . in rest of the paper we present the results corresponds to _
u _ = 4 ev . these value of _ u _ and _ j _ are also in consistent with our earlier work@xcite .
dm for these values of _ u _ and _ j _ are calculated within the muffin - tin sphere of v atoms .
the muffin - tin sphere radii used in the calculations are 2.0 , 2.0 , 1.5 , 1.8 and 1.5 bohr for zn , cd , mg , v and o , respectively .
( 6,6,6 ) k - point mesh size are used .
convergence target of total energy was achieved below 10@xmath17 hartrees / cell .
it is important to note that the values of dm show small dependence on the size of the muffin - tin radii used in the calculation .
however the small variation in the dm due to change in the values of @xmath18 and muffin - tin radii is not expected to change in the oo pattern obtained in the present work .
in the context of geometrical frustration the basic structural unit of the spinel vanadates is the tetrahedron shown in fig .
1 , where four v atoms represented by v1 , v2 , v3 and v4 are sitting at the four corners of the tetrahedron . in order to know the orbital occupancies of these v atoms we have calculated the dm of these atoms given in spherical harmonic representation . the dm of v1 and v3 ( v2 and v4 ) are found to be the same suggesting the same orbital occupancies for v1 and v3 ( v2 and v4 ) atoms .
if one compares the dm of v1 and v2 atoms one can see the obvious differences in the up - spin channel as given in tables 13 for znv@xmath15o@xmath16 .
the dm for the combination of @xmath19 , @xmath20 element having values ( @xmath142,0 ) , ( 0,@xmath142 ) , ( -1,1 ) , ( 1,-1 ) of v@xmath12 and v@xmath0 ( in brackets ) atoms are different . for example ( -1,1 ) element of dm for v1 and v2 atoms are 0.32 and -0.32 , respectively .
this suggests the different orbital occupancies for v1 and v2 atoms .
this is also valid for mgv@xmath15o@xmath16 and cdv@xmath15o@xmath16 compounds as evident from tables 49 .
the dm for the combination of @xmath19 , @xmath20 element having values ( -1,1 ) , ( 1,-1 ) of v@xmath12 and v@xmath0 ( in brackets ) atoms for mgv@xmath15o@xmath16 and ( @xmath142,0 ) , ( 0,@xmath142 ) , ( -1,1 ) , ( 1,-1 ) of v@xmath12 and v@xmath0 ( in brackets ) atoms for cdv@xmath15o@xmath16 compounds are different .
this also suggests the different orbital occupancies for v1 and v2 atoms . from the above observations one can conclude the different orbital occupancies for v1 and v2 atoms for all the three compounds .
normally , the oo is represented in the cubic harmonics . in order to know the orbital occupancies of v atoms in these compounds , we have considered the only those elements of the dm whose magnitude is @xmath21 0.1 . from the given dm in table 1
it is not obvious to judge which orbitals are occupied in the cubic harmonic representation . in order to get the exact orbital occupancies of the strongly correlated electron system in the cubic harmonics for the v1 and v2 atoms
we have used the following method : the electronic charge distribution of the strongly correlated electron systems in the spherical polar coordinate system can be expressed by using formula , @xmath22 where @xmath23 and n@xmath24 is a spherical harmonic function and @xmath25 , @xmath26 element of the dm , respectively@xcite .
normally strongly correlated orbitals like 3d and 4f are responsible for the oo in the strongly correlated compounds . for 3@xmath27 electrons system like spinel vanadates where only up spin electrons are responsible for the oo one can write eqn .
1 as @xmath28 for _ d _ orbitals there will be five different values for the @xmath29 corresponding to each values of m@xcite . these are @xmath30 = @xmath31@xmath32@xmath33 , @xmath34=@xmath35@xmath36@xmath37 and @xmath38=@xmath31@xmath39 .
putting these values in eqn .
2 and using relations : @xmath40 = @xmath41@xmath42@xmath43 , @xmath44 = @xmath41@xmath42@xmath45 and @xmath46 = @xmath41@xmath47 , one can find the electron densities of vanadium atoms in the cartesian coordinate .
+ + by using the above given procedure we can calculate the electron densities n@xmath12 and n@xmath0 of v1 and v2 atoms , respectively in cartesian coordinate which are @xmath48\\ & = \frac{15}{32\pi(x^{2}+y^{2}+z^{2})^{2}}[0.86(x^{4}+y^{4}+2x^{2}y^{2})+0.56(x^{4}+y^{4}-6x^{2}y^{2})+3.28(x^{2}+y^{2})z^{2}\\ & -2.56(y^{2}-x^{2})z^{2}+0.44z^{4}+0.11(x^{4}+y^{4})+0.22x^{2}y^{2}-0.44(x^{2}+y^{2})z^{2}]\\ & = \frac{0.15}{(x^{2}+y^{2}+z^{2})^{2}}[1.53(x^{4}+y^{4})-1.42x^{2}y^{2}+0.28(x^{2}+y^{2})z^{2}+0.44z^{4}+5.12x^{2}z^{2}]\\\end{aligned}\ ] ] thus @xmath49\end{aligned}\ ] ] and @xmath50\\ & = \frac{15}{32\pi(x^{2}+y^{2}+z^{2})^{2}}[0.86(x^{4}+y^{4}+2x^{2}y^{2})+0.56(x^{4}+y^{4}-6x^{2}y^{2})+3.28(x^{2}+y^{2})z^{2}\\ & -2.56(x^{2}-y^{2})z^{2}+0.44z^{4}+0.11(x^{4}+y^{4})+0.22x^{2}y^{2}-0.44(x^{2}+y^{2})z^{2}]\\ & = \frac{0.15}{(x^{2}+y^{2}+z^{2})^{2}}[1.53(x^{4}+y^{4})-1.42x^{2}y^{2}+0.28(x^{2}+y^{2})z^{2}+0.44z^{4}+5.12y^{2}z^{2}]\\\end{aligned}\ ] ] thus @xmath51\end{aligned}\ ] ] eqns . 3 and 4 provides the spatial distribution of the electron density in the cartesian coordinate which gives the information about the oo in znv@xmath15o@xmath16 compound .
the first four terms of eqns .
( 3 ) and ( 4 ) are same and hence contribute equal electron density in the space for both the atoms .
the last term is different and can be considered as a signature of oo as for v1 atom electrons are mainly occupied in the @xmath52 plane whereas for the v2 atom electrons are mainly occupied in @xmath53 plane .
this suggests that the @xmath3 and @xmath4 orbitals are mainly occupied by v1 and v2 atoms , respectively which corresponds to the anti - ferro oo in znv@xmath15o@xmath16 .
the oo of znv2o4 is studied by many researchers .
it will be interesting to compare our result with the results of other groups . based on model calculations , tsunetsugu _ et al . _ have predicted @xmath3 and @xmath4 oo .
however , the _ ab inito _ electronic structure calculations of maitra _
et al . _ have shown the existence of @xmath3+@xmath4 and @xmath3-@xmath4 oo .
it is important to note that coordinate systems used by these groups are different from our coordinate system .
these groups has set @xmath40 and @xmath44 axes along the two basal v - o bonds and @xmath46 axis along the apical v - o bond of regular vo@xmath54 octahedron . in our case
the @xmath40 and @xmath44 axes are making angles of 47.4 degree and 42.6 degree with two basal v - o bonds and @xmath46 axis is making an angle of 4.8 degree with the apical v - o bond .
this is due to the fact that in tetragonal phase the vo@xmath54 octahedron is distorted one . in order to compare our result with the results of tsunetsugu _ et al . _ and maitra _
_ we have rotated the @xmath40 and @xmath44 axes by 42.6 degree and calculated the electron densities for v1 and v2 atoms in new coordinate system .
we get @xmath55\\ & = \frac{15}{32\pi(x^{2}+y^{2}+z^{2})^{2}}[0.86(x^{4}+y^{4}+2x^{2}y^{2})+2.2x^{2}y^{2}-0.55(x^{4}+y^{4}-2x^{2}y^{2})\\ & -0.35xy(y^{2}-x^{2})+3.28(x^{2}+y^{2})z^{2}+5.09xyz^{2}-0.20(y^{2}-x^{2})z^{2}+0.44z^{4}\\ & + 0.11(x^{4}+y^{4})+0.22x^{2}y^{2}-0.44(x^{2}+y^{2})z^{2}]\\ & = \frac{0.15}{(x^{2}+y^{2}+z^{2})^{2}}[0.42(x^{4}+y^{4})+5.24x^{2}y^{2}+0.44z^{4}-0.35xy(y^{2}-x^{2})+0.40x^{2}z^{2}\\ & + 2.64(x^{2}+y^{2})z^{2}+5.09xyz^{2}]\\\end{aligned}\ ] ] thus @xmath56 \end{aligned}\ ] ] and @xmath57\\ & = \frac{15}{32\pi(x^{2}+y^{2}+z^{2})^{2}}[0.86(x^{4}+y^{4}+2x^{2}y^{2})+2.2x^{2}y^{2}-0.55(x^{4}+y^{4}-2x^{2}y^{2})\\ & -0.35xy(y^{2}-x^{2})+3.28(x^{2}+y^{2})z^{2}-5.09xyz^{2}-0.20(x^{2}-y^{2})z^{2}+0.44z^{4}\\ & + 0.11(x^{4}+y^{4})+0.22x^{2}y^{2}-0.44(x^{2}+y^{2})z^{2}]\\ & = \frac{0.15}{(x^{2}+y^{2}+z^{2})^{2}}[0.42(x^{4}+y^{4})+5.24x^{2}y^{2}+0.44z^{4}-0.35xy(y^{2}-x^{2})+0.40y^{2}z^{2}\\ & + 2.64(x^{2}+y^{2})z^{2}-5.09xyz^{2}]\\\end{aligned}\ ] ] thus @xmath58 \end{aligned}\ ] ] the first four terms of eqns .
( 5 ) and ( 6 ) are same and suggesting the same electron density in the space for both the atoms . in order to find the oo pattern we have to look at the terms which are different in the above equations . on dropping the first four terms from eqns .
( 5 ) and ( 6 ) we get @xmath59\end{aligned}\ ] ] and @xmath60 \end{aligned}\ ] ] on neglecting the last two terms which are very small in comparison to first term , eqns .
( 7 ) and ( 8) clearly indicates that the electron densities of both v1 and v2 atoms are mainly in @xmath10+@xmath9 and @xmath10-@xmath9 planes , respectively .
hence @xmath3+@xmath4 and @xmath3-@xmath4 orbitals are mainly occupied at the neighboring sites .
this type of orbital occupancies corresponds the anti - ferro oo in znv@xmath15o@xmath16 , which is similar to that shown by maitra _
+ + by using table 5 and following the same procedure as in the case of znv@xmath0o@xmath1 we can get electron densities of v1 and v2 atoms for mgv@xmath0o@xmath1 . the electron densities of v1 and v2 atoms are @xmath61\end{aligned}\ ] ] and @xmath62\end{aligned}\ ] ] respectively .
it is clear from these equations that first four terms are same and hence provide same electronic distributions in the space .
last term is different which shows that the electronic distribution for v1 and v2 atoms are in @xmath53 and @xmath52 planes , respectively . suggesting the occupations of @xmath4 and @xmath3 orbitals in the global coordinate system .
in mgv@xmath0o@xmath1 the @xmath40 and @xmath44 axes are making angles of 47.67 degree and 43.33 degree with two basal v - o bonds and @xmath46 axis is making an angle of 5.7 degree with the apical v - o bond .
on rotating the @xmath40 and @xmath44 axes by 43.33 degree , the calculated electron densities for v1 and v2 atoms in local octahedral coordinate system become @xmath63\end{aligned}\ ] ] and @xmath64 \end{aligned}\ ] ] respectively .
the first four terms of eqns . ( 11 ) and ( 12 ) are same and suggesting the same electron density in the space . on dropping the first four terms from eqns .
( 11 ) and ( 12 ) we get @xmath65\end{aligned}\ ] ] and @xmath66 \end{aligned}\ ] ] on neglecting the last two terms which are almost four times smaller than the first term we get the same oo pattern for mgv@xmath0o@xmath1 compound in the local octahedral coordinate system as observed in the znv@xmath0o@xmath1 compound .
the small but different contribution from the last two terms are coming due to the fact that the vo@xmath54 octahedron is not a regular one .
+ + by using table 8 and applying the same procedure as mentioned above , the calculated electron densities of v1 and v2 atoms , respectively for cdv@xmath0o@xmath1 are given by @xmath67\end{aligned}\ ] ] and @xmath68\end{aligned}\ ] ] here also the first four terms in eqns .
( 15 ) and ( 16 ) are same and last term is different .
these equations also show the occupation of @xmath3 and @xmath4 orbitals in the global coordinate system as seen for the znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 compounds . in cdv@xmath0o@xmath1 the @xmath40 and @xmath44 axes are making angles of 40.89 degree and 49.11 degree with two basal v - o bonds and
@xmath46 axis is making an angle of 6.1 degree with the apical v - o bond .
when we rotate the @xmath40 and @xmath44 axes by 40.89 degree , then the calculated electron densities for v1 and v2 atoms in local octahedral coordinate system become @xmath69\end{aligned}\ ] ] and @xmath70\end{aligned}\ ] ] respectively . here
also the first four terms in eqns .
( 17 ) and ( 18 ) are same .
hence by dropping these terms we get @xmath71\end{aligned}\ ] ] and @xmath72\end{aligned}\ ] ] these equations also suggest the similar oo pattern for cdv@xmath0o@xmath1 compound in the local octahedral coordinate system as seen in the znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 compounds .
thus our studies on all the three compounds show exactly the same oo for all the compounds in the global coordinate system where @xmath3 and @xmath4 orbitals are mainly occupied by the neighboring v atoms as shown in the fig .
2 . in the local octahedral coordinate system
the oo patterns for all the compounds show small deviation from @xmath3+@xmath4 and @xmath3-@xmath4 oo due to the presence of distorted vo@xmath54 octahedron in the tetragonal structure of these compounds .
such deviation is expected to be important in understanding the magnetic properties of the compounds as the oo of a compound also decides the spin - ordering . finally , it is important to note that quantification of the oo by including soc will be an interesting studies on these compounds as the soc is found to play an important role in deciding the magnetic properties of the compounds .
the study of the effect of soc on the oo is under progress and will be reported elsewhere .
in conclusion we have studied the orbital ordering ( oo ) of the spinel vanadates av@xmath0o@xmath1 ( a @xmath2 zn , cd and mg ) , which is considered to be responsible for the low temperature structural transition from cubic to tetragonal phase in these compounds , by using density matrices approach to the oo .
the gga+u based _ ab initio _ density functional theory has been used to calculate the density matrices of the vanadium atoms . in order to avoid extra complications we have not included the spin - orbit coupling for the valance electrons . in global ( local octahedral )
coordinate system anti - ferro orbital ordering has been observed in these compounds , where @xmath3 and @xmath4 ( @xmath3+@xmath4 and @xmath3-@xmath4 ) orbitals are mainly occupied at the neighboring v sites .
.the density matrices correspond to 3@xmath5 up electrons of v@xmath12 and v@xmath0 ( in brackets ) atoms for znv@xmath0o@xmath1 when _ u _ = 3 ev and _ j _ = 0.5 ev . [ cols="^,^,^,^,^,^,^",options="header " , ] | in this work we apply the density matrices approach to orbital ordering ( oo ) in order to study the oo of the spinel vanadates av@xmath0o@xmath1 ( a @xmath2 zn , cd and mg ) , which is normally believed to be responsible for the structural transition from cubic to tetragonal phase observed in these compounds .
the density matrices of vanadium atoms are obtained by using _ state - of - the - art _ full - potential linearized augmented plane wave method based gga+u calculations . in the absence of spin - orbit coupling ,
the present study shows the existence of anti - ferro oo in the global ( local octahedral ) coordinate system where @xmath3 and @xmath4 ( @xmath3+@xmath4 and @xmath3-@xmath4 ) orbitals are mainly occupied at the neighboring v sites for all the compounds . |
we are concerned here with proving localization properties of multi - dimensional continuum random schrdinger operators in the fluctuation boundary regime .
such results were first found via the method of multiscale analysis , which had been developed in the 80s to handle lattice models and was later extended to the continuum ( for a rather complete history and list of references on multiscale analysis see @xcite and for some of the more recent developments @xcite ) .
later , the fractional moment method was developed @xcite as an alternative approach to the same problem , also initially for lattice models .
it leads to a stronger form of dynamical localization than multiscale analysis ( see @xcite ) and has provided much shorter and more transparent proofs in the lattice case , for example @xcite .
it was recently shown in @xcite that all the main features of the fractional moment approach also apply to continuum random schrdinger operators .
this extension required substantial new input from operator theory and harmonic analysis .
the paper @xcite provides a framework of necessary and sufficient criteria for localization in terms of fractional moment bounds , which can be verified for a rather broad range of regimes .
one of our goals here is to complement the general framework from @xcite by focusing exclusively on presenting a short and self - contained proof of localization properties via fractional moments for one specific regime , where the technical effort remains minimal .
for this we pick a fairly general setting we label the _ fluctuation boundary regime_. this is described by a random schrdinger operator of anderson - type in @xmath0 , where our approach allows for quite arbitrary background potentials and geometries of the random impurities , provided the ground state energy is induced by rare events ( fluctuations ) and therefore sensitive to changes in the random parameters . the goal is to prove localization in the vicinity of the bottom of the spectrum . of course
, various versions of the fluctuation boundary regime have been studied in many works and we borrowed the term from @xcite .
another motivation for our work is that we want to extend the fractional moment method to situations where the random potential does not satisfy a covering condition , i.e. where the individual impurity potentials have small supports which do not cover all of @xmath1 .
this condition , which was required for the technical approach to the continuum found in @xcite , is not natural in the fluctuation boundary regime and should not be needed there as has already been verified via multiscale analysis .
particularly interesting examples are random surface potentials which act in a small portion of space only .
nevertheless , they lead to a fluctuation boundary by creating new `` surface spectrum '' below the `` bulk spectrum '' . in our main result ,
theorem [ t1 ] below , the fluctuation boundary regime will be described in form of an abstract condition . for random surface potentials , which will be discussed as an application
, this condition follows in an appropriate setting from a result proven in @xcite in order to derive lifshits tails . another application concerns models with additional random displacements as were originally studied in @xcite . let us confess that we require absolutely continuous distribution of random couplings .
while it might be possible to relax this to hlder continuous distribution ( as has been done in the lattice case , e.g. @xcite ) , the fractional moment method is so far less flexible in that respect than the multiscale technique . in particular , see the variant of multiscale analysis adapted to bernoulli - anderson models recently developed in @xcite and an application of similar ideas to poisson models announced in @xcite .
let us now describe our results in more detail after introducing some notation : on @xmath1 we often consider the supremum norm @xmath2 and write @xmath3 for the @xmath4-dimensional cube with sidelength @xmath5 centered at @xmath6 .
for an open set @xmath7 we denote the restriction of the schrdinger operator @xmath8 to @xmath9 with dirichlet boundary conditions by @xmath10 . in our results
we assume @xmath11 and rely upon the following assumptions , which guarantee self - adjointness and lower semi - boundedness of all the schrdinger operators appearing in this paper : 1 .
the background potential @xmath12 is real - valued , @xmath13 .
2 . the set @xmath14 , where the random impurities are located , is uniformly discrete , i.e. , @xmath15 .
3 . the random couplings @xmath16 , @xmath17 , are independent random variables supported in @xmath18 $ ] for some @xmath19 and with absolutely continuous distribution of bounded density @xmath20 with a uniform bound @xmath21 . + the single site potentials @xmath22 , @xmath17 satisfy @xmath23 for all @xmath24 with @xmath25 independent of @xmath26 . @xmath27 and
@xmath28 the most important condition expresses the fact that the ground state energy comes from those realizations of the potential that vanish on large sets : 1 .
denote @xmath29 and let @xmath30 the subscript @xmath31 standing for full coupling .
+ assume that @xmath32 is a _ fluctuation boundary _ in the sense that a. @xmath33 , and b. there is @xmath34 and @xmath35 such that for @xmath36 , all @xmath37 and @xmath38 @xmath39 \not=\emptyset\bigr)\leq l^{-m_d } .\ ] ] by @xmath40 we denote the characteristic function of the unit cube centered at @xmath6 . in the following it
is understood that @xmath41 if @xmath42 or @xmath43 have measure zero .
our main result is [ t1 ] let @xmath11 and assume _
( a1)-(a4)_. then there exist @xmath44 , @xmath45 , @xmath46 and @xmath47 such that for @xmath48 , all open sets @xmath49 and @xmath50 , @xmath51 exponential decay of fractional moments of the resolvent as described by implies spectral and dynamical localization in the following sense : [ t2 ] let @xmath11 , assume _ ( a1)-(a4 ) _ and let @xmath52 be given as in theorem [ t1 ] .
then : 1 . for all open sets
@xmath53 the spectrum of @xmath10 in @xmath52 is almost surely pure point with exponentially decaying eigenfunctions .
2 . there are @xmath54 and @xmath47 such that for all @xmath55 and open @xmath53 , @xmath56 where the supremum is taken over all borel measurable functions @xmath57 which satisfy @xmath58 pointwise and @xmath59 is the spectral projection for @xmath10 onto @xmath52 .
dynamical localization should be considered as the special case @xmath60 in ( b ) , with the supremum taken over @xmath61 .
the proof of theorem [ t1 ] is given in section [ sec : mainproofs ] .
this will be done by a self - contained presentation of a new version of the continuum fractional moment method . while we use many of the same ideas as @xcite , due to the lack of a covering condition
we can not rely any more on the concept of averaging over local environments " , heavily exploited in @xcite .
it is interesting to note that , in some sense , we instead use a global averaging procedure .
technically , this actually leads to some simplifications compared to the method in @xcite , as repeated commutator arguments can be replaced by simpler iterated resolvent identities .
we also mention that exponential decay in will follow from smallness of the fractional moments at a suitable initial length scale ( the localization length ) via an abstract contraction property . as technical tools we need combes - thomas bounds ( in operator norm as well as in hilbert - schmidt norm ) and a weak-@xmath62-type bound for the boundary values of resolvents of maximally dissipative operators , which is based on results from @xcite and was also central to the argument in @xcite .
we collect these tools in an appendix .
that theorem [ t2 ] follows from theorem [ t1 ] was essentially shown in @xcite , section 2 . in section [ sec : prooft2 ] below we will briefly discuss the changes which arise due to our somewhat different set - up .
in particular , the argument in @xcite for proving uses the covering condition @xmath63 in one occasion .
but this is easily circumvented . in sections [ sec : surface ] and
[ sec : displacement ] we apply our main result to concrete models by verifying assumption ( a4 ) for these models . in section [ sec : surface ]
we consider anderson - type random potentials supported in the vicinity of a lower - dimensional surface .
the `` usual '' fully stationary anderson model is considered in section [ sec : displacement ] .
the fact that we do nt have to assume a covering condition leads to high flexibility in the geometry of the random scatterers . we could use this to go for far reaching generalizations of anderson models . instead , we restrict ourselves to the treatment of additional random displacements as was done in @xcite .
we could have extended theorem [ t1 ] in at least two different ways , but refrained from doing so to keep the proofs as transparent as possible : a. the restriction to @xmath11 is not necessary .
we use it because in this case the abstract fractional moment bound in corollary [ cor : fracmoments ] is more directly applicable to our proof of theorem [ t1 ] than in higher dimension ( which technically can be traced back to the fact that @xmath64 is a hilbert - schmidt operator only for @xmath11 ) . in higher dimension
more iterations of resolvent identities would be needed to yield the hilbert - schmidt multipliers required by corollary [ cor : fracmoments ] , leading to more involved summations in the arguments of section [ sec : mainproofs ] .
b. instead of bounded @xmath65 we can work with relatively @xmath66-bounded @xmath65 , i.e. allow for suitable @xmath67-type singularities in the single site potentials .
in the course of our proofs they could be absorbed " into resolvents using standard arguments from relative perturbation theory . in principle , our arguments can also be used to prove localization at fluctuation type band edges more general than the bottom of the spectrum without using a covering condition as in @xcite . but this would require to be much more specific with settings and assumptions and , in particular , with the geometry of the impurity set .
inconvenience would also arise from having to work with boundary conditions other than dirichlet .
we mention that the applications in section [ sec : surface ] improve the results on continuum random surface potentials of @xcite , obtained through the use of multiscale analysis : a. the exponentially decaying correlations of the time evolution , shown as a special case of theorem [ t2](b ) , are stronger than the dynamical bounds which follow from multiscale analysis .
b. due to the use of the recent result of @xcite on lifshitz tails for surface potentials , we do not need a condition on the smallness of the distribution of the @xmath16 near the minimum of their support as in @xcite , a progress that had been achieved in @xcite . c. we can allow for more flexibility concerning the geometry of the scatterers .
of course , due to using fractional moments we can not include single site measures as singular as the ones considered in @xcite but instead have to assume absolute continuity of the @xmath68 .
this section is entirely devoted to the proof of theorem 1 . for a convenient normalization write @xmath69^{{\mathcal{i}}},\end{aligned}\ ] ] and denote the product measure @xmath70 on @xmath71 by @xmath72 .
we write @xmath73 note that @xmath74 and that @xmath75 fixing an open set @xmath76 we write @xmath77 whenever @xmath78 . since @xmath79 due to our choice of dirichlet boundary conditions , and @xmath80 we know that @xmath81 . the resolvent equation yields @xmath82 an identity that will be used over and again .
the other workhorse result is the following averaging estimate , that follows from corollary [ cor : fracmoments ] in the appendix below , taking into account the uniform boundedness of the densities @xmath20 .
[ workhorse ] for all @xmath83 there is @xmath84 such that @xmath85 as a warm - up , we prove boundedness of fractional moments : [ l3 ] let @xmath86 , @xmath87 $ ] and @xmath88 .
then @xmath89 we use above and write , suppressing the superscript @xmath90 and the subscript @xmath91 mostly : @xmath92 the first two terms on the r.h.s . of this equation obey an exponential bound due to the combes - thomas estimate , see subsection [ ct ] below : @xmath93 and @xmath94 with @xmath95 and @xmath96 depending on @xmath97 only . in the last estimate we have used that @xmath98 is uniformly discrete . for the third term , expand @xmath99 and use the boundedness of the @xmath100 and the fact that @xmath101 to estimate @xmath102 we now fix @xmath103 and use the workhorse lemma [ workhorse ] to conclude @xmath104 and since @xmath105 where @xmath106 majorizes the size of the support of @xmath22 according to assumption ( a3 ) . note that here and in the following we use the convention that @xmath107 , @xmath84 , etc . denote constants that only depend on non - crucial quantities and may change from line to line . in particular , the constants are independent of @xmath108 and the random background .
now , we can sum up the last terms and get the assertion .
\(i ) in this proof it is still quite easy to see how to extend to arbitrary dimension through iterations of the resolvent identity .
it will be harder to keep track of this later .
\(ii ) note that due to the @xmath109-summations , averaging over the @xmath16 is required for all @xmath26 , i.e. , is global . in @xcite , due to the covering condition , an argument is provided that only requires averaging over local environments of @xmath6 and @xmath110 and proves lemma [ l3 ] for arbitrary finite intervals @xmath87 $ ] , i.e. without requiring @xmath86 .
\(iii ) the above proof shows that also holds in hs - norm , but this will not be used below .
we will now start an iterative procedure that will show exponential decay of @xmath111 in @xmath112 for energies sufficiently close to @xmath32 .
clearly , it suffices to consider @xmath113 . in view of the preceding lemma
the following quantity is finite : @xmath114 moreover , we should actually keep in mind the dependence on the interval @xmath115 $ ] . in fact , @xmath97 will later be chosen small enough . in order to use
that @xmath32 appears rarely as an eigenvalue for boxes of side length @xmath116 we exploit the resolvent identity and what is sometimes called the simon - lieb inequality in a way visualized in figure 1 ! consider @xmath117 furthermore , with @xmath106 as in assumption ( a3 ) , define @xmath118 the geometry is chosen in such a way that @xmath119 and @xmath120 are stochastically independent . for @xmath119 we can use the fluctuation boundary assumption to get small fractional moments and the right size of @xmath116 will be adjusted .
but all that later ... thus , by the simon - lieb inequality ( e.g. @xcite , sect .
2.5 ) @xmath121 where @xmath122 only depends on @xmath123 and the interval @xmath52 .
the basic idea for proving exponential decay of @xmath124 is to establish a recurrence inequality for energies sufficiently close to @xmath32 .
this recurrence inequality is described in proposition [ master ] below and allows to apply a discrete gronwall - type argument found in lemma [ lem.estimate.operator.a ] below . to this end
we exploit smallness of fractional moments of the first factor on the r.h.s . of ( sli ) for energies close to @xmath32 and sufficiently large , but fixed , @xmath116 : this will follow from ( a4)(ii ) as is presented in the following lemma [ startup ] .
fractional moments of the second factor are bounded due to lemma [ l3 ] ( up to a polynomial factor in @xmath116 ) .
finally , we use the third factor to start an iteration ( with @xmath6 replaced by sites @xmath125 covering the layer @xmath126 ) . by construction , the first and third factor on the r.h.s . of ( sli )
are probabilistically independent .
unfortunately , the second factor introduces a correlation which prevents us from simply factoring the expectation . we will rely on a version of the re - sampling procedure developed in @xcite to solve this problem .
moreover , we will not use lemma [ l3 ] , but apply lemma [ workhorse ] directly to bound certain conditional expectations .
this will result in proposition [ master ] below . [ startup ] for @xmath127 as in _ ( a4 ) _ and @xmath128 there is @xmath129 such that for all @xmath37 , open @xmath130 , @xmath131 $ ] , @xmath132 and @xmath133 with @xmath134 we have @xmath135 where @xmath136 .
divide @xmath71 into the good and bad sets @xmath137 since @xmath138 by our choice of dirichlet boundary conditions , ( a4 ) implies that @xmath139 we split the expectation into contributions from the good and bad sets : by the improved combes - thomas bound subsection [ ct ] we get , for @xmath140 , @xmath141 : @xmath142 this gives a uniform bound of the same type for the expectation over @xmath143 . for the bad set hlder with @xmath144
gives @xmath145 now we choose @xmath146 so that @xmath147 if @xmath148 .
putting things together , we get @xmath149 if @xmath116 is large enough we can use @xmath150 and @xmath151 to see that the r.h.s .
is bounded as asserted .
the exponential decay of the @xmath124 will follow from the following result , whose proof will take most of the present section .
[ master ] there exist @xmath35 , @xmath152 , @xmath153 and @xmath154 , all depending on @xmath155 , such that for @xmath37 and @xmath156 $ ] the above defined @xmath124 satisfy : @xmath157 we now restrict to the energy interval @xmath158 $ ] assuming @xmath116 is large enough to guarantee that @xmath159 . using ( sli ) above and denoting @xmath160 we get that @xmath161 note that @xmath162 and @xmath163 are stochastically independent .
unfortunately , they are correlated via @xmath164 .
fix @xmath165 to estimate @xmath166 . using the preceding lemma
, we get that @xmath167 for @xmath116 large enough .
we get that @xmath168 we can now expand @xmath169 to split off a uniformly bounded ( in @xmath170 ) term : @xmath171 since @xmath172 we have that @xmath173 is uniformly bounded .
thus , we get @xmath174 as @xmath162 and @xmath163 are independent . here
@xmath175 expanding @xmath176 we get , for some @xmath153 , that @xmath177 whence @xmath178 to estimate @xmath179 we begin by expanding @xmath180 since @xmath52 has positive distance from @xmath181 , we have the combes - thomas bound @xmath182 for the norm of the first two terms on the r.h.s . of ,
see appendix [ ct ] . here
@xmath47 and @xmath183 are uniform in the randomness , @xmath184 , @xmath185 and @xmath186 . expanding the third term and using boundedness of the @xmath187 s yields @xmath188 setting @xmath189 , and only summing over those @xmath190 for which the corresponding @xmath191-terms touch @xmath192 .
a similar argument applied to @xmath193 leads to @xmath194 where we have chosen @xmath195 .
finally , expand @xmath196 where @xmath197 . combining all this
we get that @xmath198 the most complicated of these terms is the last one ; it will be obvious how to estimate the first three once we have established a bound for the last one according to the assertion of proposition [ master ] .
thus we have to estimate @xmath199 where @xmath200 if it were nt for the @xmath201-terms , the @xmath202 and @xmath203 would be independent , leading to an estimate like in above .
we will reinforce a certain kind of independence through _ re - sampling_. for fixed @xmath204 we introduce new independent random variables @xmath205 , @xmath206 , independent of the @xmath207 , @xmath208 , and with the same distribution as the @xmath207 .
we denote the corresponding space by @xmath209 , the corresponding probability by @xmath210 and the expectation with respect to @xmath210 by @xmath211 .
consider @xmath212 and note that @xmath213 does nt depend on the @xmath214 , @xmath215 .
the resolvent identity for @xmath216 gives @xmath217 we insert this for @xmath202 and @xmath203 and get @xmath218 and , similarly , @xmath219 now we can estimate @xmath220 .\ ] ] this gives a sum of four terms we have to control .
let s start with the easiest one @xmath221 .\ ] ] denote @xmath222 for a random variable on @xmath223 , so that @xmath224 is nothing but the conditional expectation with respect to the @xmath225-field generated by the family @xmath226 . according to the usual rules for conditional expectations
: @xmath227 \notag \\ & = \operatorname{\widehat{{\mathbb{e}}}}\operatorname{{\mathbb{e}}}\left [ \| \widehat{t}_{\beta,\gamma}\|^s\|\widehat{q}_{\beta',\gamma'}\|^s \operatorname{{\mathbb{e}}}(\| s_{\alpha,\alpha ' } \|^s|\alpha,\alpha')\right]\end{aligned}\ ] ] since the @xmath228 and @xmath229 are independent of @xmath230 .
using the workhorse lemma [ workhorse ] and the combes - thomas estimate proposition [ prop : cths ] we get @xmath231 term comes from covering @xmath232 and @xmath233 .
we have @xmath234 = \operatorname{{\mathbb{e}}}\left [ \| { t_{\beta,\gamma}}\|^s\|{q_{\beta',\gamma'}}\|^s\right ] = \operatorname{{\mathbb{e}}}\left [ \| { t_{\beta,\gamma}}\|^s\right]\operatorname{{\mathbb{e}}}\left [ \|{q_{\beta',\gamma'}}\|^s\right]\ ] ] since the @xmath235 s have the same distribution as the @xmath187 s and the @xmath236 s and @xmath237 s are independent . inserting into gives @xmath238\operatorname{{\mathbb{e}}}\left [ \|{q_{\beta',\gamma'}}\|^s\right].\ ] ] we will now treat the latter two terms separately : denote by @xmath239 those lattice points whose 1-cubes support @xmath240 . by combes - thomas once again : @xmath241 by assumption on the size of the support of @xmath240 we see that @xmath242 is uniformly bounded . this and uniform discreteness of @xmath98 gives @xmath243 for the @xmath202-term we have @xmath244 if @xmath245 , lemma [ startup ] gives @xmath246 if , on the other hand , @xmath247 then @xmath248 or @xmath249 , so that the uniform bound of lemma [ l3 ] for @xmath250 together with the combes - thomas bound for @xmath251 , resp .
@xmath252 gives , for @xmath116 large enough , @xmath253 combined we get that , again for @xmath116 sufficiently large , @xmath254 where an extra factor @xmath255 arises through the number of terms considered .
joining step 1 , step 2 and the bound @xmath256 which is a contribution to @xmath257 ( and therefore @xmath179 ) bounded by one of the type asserted in proposition [ master ] . a look back at shows that we still have to estimate three terms similar to @xmath258 of which the last one , @xmath259\ ] ] is the most complicated one . using steps 1 and 2 above as well as the steps below it will be clear how to treat the two remaining terms . we start taking the conditional expectation : @xmath260\\ & \leq \operatorname{\widehat{{\mathbb{e}}}}\operatorname{{\mathbb{e}}}\bigl\lbrack\operatorname{{\mathbb{e}}}(\| \widetilde{t}_{\beta,\gamma}\|^{3s } |\alpha,\alpha ' , \gamma,\gamma')^\frac13\\ & \qquad\quad \cdot\operatorname{{\mathbb{e}}}(\| s_{\alpha,\alpha ' } \|^{3s } |\alpha,\alpha ' , \gamma,\gamma')^\frac13\cdot\operatorname{{\mathbb{e}}}(\|\widetilde{q}_{\beta',\gamma'}\|^{3s } |\alpha,\alpha ' , \gamma,\gamma')^\frac13\bigr\rbrack\end{aligned}\ ] ] by hlder s inequality .
like above , the middle term can , up to @xmath261 , be estimated by @xmath262 recall that @xmath263 where @xmath264 can be excluded from the summation as @xmath265 does nt touch @xmath266 .
integration over @xmath214 and @xmath267 gives a uniform bound by the workhorse lemma [ workhorse ] : @xmath268 so that , as the sum has only three terms , @xmath269 similarly , @xmath270 now @xmath271 and @xmath272 are independent so that @xmath273 \cdot\operatorname{\widehat{{\mathbb{e}}}}\operatorname{{\mathbb{e}}}[\sigma_q ] \cdot e_{\alpha,\alpha ' } .\ ] ] since the @xmath214 and the @xmath205 have the same distribution , we can omit the hats in @xmath274 and @xmath275 and replace @xmath276 by @xmath277 in the bounds for @xmath278 $ ] and @xmath279 $ ] to be derived below .
we start with the @xmath237-term .
combes - thomas , proposition [ prop : cths ] gives @xmath280 this will be used to deal with the term for @xmath281 which appears in the sum over @xmath282 ; since @xmath283 we get that : @xmath284 for the terms @xmath285 and @xmath286 in the sum we borrow from @xmath287 above and use that @xmath288 if @xmath289 and @xmath290 : @xmath291 summing each of the three contributions from and to @xmath292 over @xmath293 ( and extending the @xmath125-sum in to all of @xmath294 ) gives @xmath295 \leq c l^{2(d-1)}\sum_{x',y'\in{\mathbb{z}}^d } { \mathrm{e}}^{-\frac{c}{l}|x - x'|-\mu_1 s we now show that summation over @xmath296 gives a small prefactor . by exponential decay : @xmath297 we analyze @xmath298 if @xmath299
then either @xmath300 or @xmath301 . since @xmath302 @xmath303 is bounded by @xmath304
; see step 2 above . if , on the other hand @xmath305 we can use lemma [ startup ] above to estimate @xmath306 summing up these terms we get that @xmath307 \leq c l^{3d-1-\frac12 m_d}\ ] ] since @xmath308 run through at most @xmath309 different points of @xmath98 in @xmath192 . putting the estimates from ( [ q]),([t]),([s ] ) together we arrive at : @xmath310 which is the desired bound . to deal with the other terms appearing in @xmath311
we just combine the corresponding steps to control the @xmath236 and @xmath237-sums respectively .
this concludes the proof of proposition [ master ] . for energies sufficiently close to @xmath32 we will now complete the proof of exponential decay of @xmath124 , and thus of theorem [ t1 ] , by applying a discrete gronwall - type argument to the recursion inequality established in proposition [ master ] .
for @xmath46 consider the weighted @xmath312-space @xmath313 i.e. , for @xmath314 , @xmath315 [ lem.estimate.operator.a ] the operator @xmath316 defined by @xmath317 is bounded as an operator in @xmath318 as well as an operator in @xmath319 with @xmath320 the norm of @xmath316 in @xmath318 is the same as the norm of the operator @xmath321 in @xmath319 with kernel @xmath322 thus @xmath323 with the substitutions @xmath324 , @xmath325 , @xmath326
. bound the latter exponent through @xmath327 after cancellation the integral factorizes and gives for @xmath328 after scaling .
the bound for @xmath329 is found more directly .
this may be applied to the situation of proposition [ master ] as it shows that for @xmath116 sufficiently large the operator @xmath316 with kernel @xmath330 has norm less than one , both as an operator in @xmath331 and an operator in @xmath319 .
fix this @xmath116 and choose @xmath332 , @xmath333 $ ] in theorem [ t1 ] and the definition of @xmath124 .
the recursion inequality now takes the form @xmath334 with @xmath335 .
the conclusion of the proof of theorem [ t1 ] is now the content of [ lem.t.in.x ] @xmath336 with @xmath337 define the diagonal operator @xmath338 which is an isometry from @xmath318 to @xmath319 .
let @xmath339 and @xmath340 .
let @xmath341 .
then implies that componentwise @xmath342 since @xmath343 is bounded and @xmath316 a bounded operator in @xmath319 , we have that @xmath344 and @xmath321 is a bounded operator in @xmath345 with non - negative kernel . thus we obtain from that @xmath346 holds with finite components .
summation yields @xmath347 and thus @xmath348 for all @xmath349 .
@xmath350 is a contraction and @xmath351 .
thus @xmath352 in @xmath312 and componentwise .
also , @xmath353 is a contraction and @xmath354 .
thus @xmath355 and componentwise as @xmath356 .
we conclude @xmath357 lemma [ lem.t.in.x ] is proved .
that the localization properties stated in theorem [ t2 ] follow from the fractional moment bound for the resolvent established in theorem [ t1 ] was demonstrated in @xcite . here
we want to comment on two minor changes in the argument which are due to our somewhat different set - up .
first we note that spectral and dynamical localization as established in parts ( a ) and ( b ) of theorem [ t2 ] hold for restrictions of @xmath8 to arbitrary open domains @xmath90 , and , in particular , that the exponential decay established in equation holds with respect to the standard distance @xmath112 rather than the domain adapted distance @xmath358 used in @xcite . given that the corresponding bound in theorem [ t1 ] is true for arbitrary @xmath90 and in standard distance , this follows with exactly the same proof as in section 2 of @xcite ( with one exception discussed below ) . that the authors of @xcite chose to work with the domain adapted distance was in order to include more general regimes in which extended surface states might exist
this is not the case in the regime considered here .
second , let us provide a few details on how to eliminate the use of the covering condition from the proof of provided in section 2 of @xcite .
as done there one first considers bounded open @xmath359 and defines @xmath360 if @xmath361 and @xmath362 are the eigenvalues and corresponding eigenfunctions of @xmath363 and @xmath364 is as above , then @xmath365 readily implies @xmath366 at this point we modify the argument of @xcite and write @xmath367 as all @xmath368 have a uniform distance from @xmath369 we get from combes - thomas proposition [ prop : ct ] that @xmath370 inserting above yields @xmath371 with @xmath372 defined as in @xcite , where the bound @xmath373 is established without any further references to the covering condition .
thus we conclude @xmath374 the rest of the proof of theorem [ t2 ] , in particular the extension of to infinite volume and a supremum over arbitrary borel functions , follows the argument in @xcite without change .
random surface models have attracted quite some interest with most of the work dealing with the discrete case @xcite and some with the continuum case @xcite , as we do here .
our aim in this section is to show that under suitable conditions such surface models obey condition ( a4 ) above . to achieve it
, we combine recent results from @xcite with a technique from @xcite . as usual ,
the background is assumed to be partially periodic : 1 .
fix @xmath375 and write @xmath376 , @xmath377 ; assume that @xmath12 is real - valued and periodic with respect to the first variable , i.e. , @xmath378 denote @xmath13 . in order to state our second requirement ,
let us recall some facts from bloch theory
. for more details , see @xcite . for @xmath379 as in ( b1 ) we get a direct integral decomposition @xmath380 where @xmath381 is the @xmath382-dimensional torus and @xmath383 with @xmath384-periodic boundary conditions on the unit strip @xmath385 .
we now fix the assumption 1 .
@xmath386 it is well known that under ( b2 ) we have that @xmath387 and there is a positive eigensolution @xmath388 of the distributional equation @xmath389 see @xcite and the references therein .
finally , our random perturbation is assumed to satisfy 1 .
the set @xmath14 , where the random impurities are located , is uniformly discrete , i.e. , @xmath390 .
moreover @xmath98 is dense near the surface @xmath391 in the sense that there exist @xmath392 such that for @xmath116 large enough and @xmath393
: @xmath394\geq c_\perp l^{d_1 } .\ ] ] we will see that ( b1)-(b3 ) ensure ( a4 ) from section 1 .
of course , there might be other ways to verify ( a4 ) for surface - like potentials so that theorems 1 and 2 could , in principle , be used for other examples .
[ t3 ] assume _
( b1)-(b3 ) _ and _ ( a3)_. then there exist @xmath44 , @xmath45 , @xmath46 and @xmath47 such that for @xmath48 , all open sets @xmath76 and @xmath50 @xmath395 in particular , the following consequences hold : 1 .
the spectrum of @xmath10 in @xmath52 is almost surely pure point with exponentially decaying eigenfunctions .
2 . there are @xmath396 and @xmath47 such that for all @xmath397 , @xmath398 the rest of this section is devoted to deducing ( a4 ) under the assumptions of the theorem .
note that this will be accomplished once we have shown the following , where @xmath399 denotes the strip of side length @xmath116 centered at @xmath400 perpendicular to the `` surface '' @xmath401 . [ p3 ] for all @xmath402 there exists @xmath403 such that for all odd integers @xmath404 and @xmath405 : @xmath406\not= \emptyset\bigr\rbrace \leq l^{-\xi } .\ ] ] in fact , ( a4)(ii ) then follows , since @xmath407 and therefore @xmath408 .
we will actually prove the analogue of proposition [ p3 ] with dirichlet boundary conditions replaced by suitable robin boundary conditions that are defined using the periodic ground state @xmath388 introduced above .
assume , for later convenience , that @xmath409 we consider on @xmath410 , @xmath411 , _ mezincescu boundary conditions _ , given as follows .
let @xmath412 where @xmath413 denotes the outer normal derivatives
. the mezincescu boundary condition can be thought of as the following requirement for functions @xmath414 in the domain of @xmath415 : @xmath416 for the formal definition of @xmath415 via quadratic forms and more background , see mezincescu s original paper @xcite as well as @xcite . in particular , we immediately get the following important relations in the sense of the corresponding quadratic forms : @xmath417 as well as @xmath418 whenever the strip @xmath410 is divided into disjoint strips @xmath419 whose closures exhaust the closure of @xmath410 . due to the form inequality ( [ eq31 ] ) above
it remains to prove the estimate for @xmath415 . denoting the bottom eigenvalue of an operator @xmath8 by @xmath420 ( _ caution _ : here our notation differs from the one in @xcite , where the second eigenvalue is denoted by @xmath420 ) we see that @xmath421\not= \emptyset \longleftrightarrow e_1(h^{s_l}_\chi)\leq e_0 + l^{-\gamma } .\ ] ] there exist @xmath422 such that @xmath423 we use here the method from @xcite .
denote @xmath424 , and its first eigenvalue by @xmath425 .
since @xmath425 increases in @xmath426 the event in ( [ eq:33 ] ) implies that @xmath425 be small for all @xmath427 which in turn implies that @xmath428 must be small .
we infer from @xcite , theorem 3.25 that the gap between the first two eigenvalues satisfies @xmath429 as in @xcite , lemma 2.3
this gives that @xmath430 now assume that @xmath431 for @xmath432 . from ( [ eq : * ] )
we get that @xmath433 with @xmath434 for @xmath435 . on the other hand @xmath436 where @xmath437 is the normalized ground state of @xmath438 .
now , the boundary condition of @xmath438 is defined so as to make sure that @xmath439 is an eigenfunction ; see the discussion in @xcite .
therefore @xmath440 and we get @xmath441 since , by ( b3 ) , there are at least @xmath442 elements of @xmath443 in @xmath444 and @xmath445 we arrive at @xmath446 with @xmath447 and independent variables @xmath68 running through an index set @xmath448 of cardinality at least @xmath449 .
if we now choose @xmath432 so small that @xmath450 where @xmath451 is smaller than the mean of all the @xmath68 s we get that : @xmath452 by a standard large deviation estimate ; see @xcite or @xcite , theorem 1.4 .
this finishes the proof of step 1 . to deduce the desired bound from step 1 we divide the strip @xmath410 into disjoint strips
@xmath419 whose closures exhaust the closure of @xmath410 and such that @xmath453 which is possible for @xmath116 large enough .
their number @xmath454 is at most @xmath455 . by ( [ eq32 ] )
we know that @xmath456 so that @xmath457 provided @xmath116 is large enough .
\(1 ) in cases where the operator @xmath8 is ergodic , a stronger bound than is provided in ( * ? ?
* proposition 5.2 ) .
their bound is in terms of the integrated density of states for which @xcite establishes lifshits asymptotics .
as we are only interested in localization properties here , the bound suffices and allows to handle the non - ergodic random potentials defined in ( b3 ) and ( a3 ) .
\(2 ) we have established localization near the bottom of the spectrum for the random surface models considered in this section .
if @xmath458 one expects for physical reasons that the entire spectrum of @xmath8 below @xmath459 ( see ( a4 ) ) is localized . a corresponding result for lattice operators has been proven in @xcite ( in situations where @xmath460 is the discrete laplacian and @xmath461 ) . to show
this for continuum models remains an open problem .
by considering the special case @xmath462 , the results of the previous section also cover `` usual '' anderson models , sometimes also called alloy models .
note that in this case ( b2 ) becomes trivial .
let us nevertheless state the assumptions and result again for this case , mainly because we want to point out below that the obtained bounds hold uniformly in the geometric parameters describing the random potential .
this will then be applied to models with random displacements . here
are the assumptions we rely upon : 1 .
@xmath12 is real - valued and periodic .
the set @xmath14 , where the random impurities are located , is uniformly discrete , i.e. , @xmath390 and uniformly dense , i.e. , there exists @xmath463 such that @xmath464 for every @xmath465 .
[ t10 ] assume _
( d1 ) , ( d2 ) _ and _ ( a3)_. then there exist @xmath44 , @xmath45 , @xmath46 and @xmath47 such that for @xmath48 , all open sets @xmath76 and @xmath50 @xmath466 in particular , the following consequences hold : 1 .
the spectrum of @xmath10 in @xmath52 is almost surely pure point with exponentially decaying eigenfunctions .
2 . there are @xmath467 and @xmath468 such that for all @xmath397 , @xmath469 here all the constants @xmath470 can be chosen to only depend on the potential through the parameters @xmath471 . to this end
we first observe that ( d1 ) , ( d2 ) and ( a3 ) imply ( a4 ) with constants @xmath472 , @xmath127 and @xmath35 only depending on the listed parameters : [ p11 ] assume _ ( d1 ) , ( d2 ) _ and _ ( a3)_. then there exist @xmath473 and @xmath474 such that 1
. @xmath475 .
2 . for @xmath36 , all @xmath37 and @xmath38 : @xmath476 \neq\emptyset\bigr)\leq l^{-m_d } .\ ] ] first we show that ( d2 ) implies that there exist @xmath477 and @xmath478 depending only on @xmath479 such that for all @xmath480 : @xmath481 the upper bound follows from uniform discreteness : @xmath482 provided @xmath483 . for the lower bound use uniform denseness : divide @xmath484 into disjoint boxes of side length @xmath485 .
if @xmath486 there are at least @xmath487 of them each of which contains at least one point from @xmath98 .
now we can use the analysis of the preceding section . since the relevant quantities depend only on the indicated parameters , the assertions follow . with this uniform version of ( a4 ) and the proofs provided in sections [ sec : mainproofs ] and [ sec : prooft2 ] we also get corresponding uniform versions of theorems [ t1 ] and [ t2 ] , i.e. theorem [ t10 ] . as a specific application of the previous observation
, we can start from an anderson model as above and additionally vary the set @xmath98 in a random way , as long as @xmath488 and @xmath485 obey uniform upper and lower bounds .
instead of formulating the most general result in this direction we look at models that were introduced in @xcite and further studied in @xcite . 1 .
let @xmath489 , @xmath490 be independent random couplings , defined on a probability space @xmath71 with distribution @xmath491 and @xmath492 as in ( a3 ) .
2 . let @xmath493 , @xmath494 be independent random vectors of length at most @xmath495 in @xmath1 ; denote the corresponding probability space by @xmath496 .
define @xmath497 assume _ ( d1 ) , ( d3 ) , ( d4)_. then , for @xmath498 as above there exist @xmath44 , @xmath45 , @xmath46 and @xmath47 such that for @xmath48 , all open sets @xmath76 and @xmath50 @xmath499 in particular , the following consequences hold : 1 .
the spectrum of @xmath10 in @xmath52 is almost surely pure point with exponentially decaying eigenfunctions .
2 . there are @xmath396 and @xmath47 such that for all @xmath397 , @xmath500 the corresponding inequality holds uniformly in @xmath501 by what we proved above .
note that in this last corollary we have not assumed that the random perturbations cover the whole space . in that respect
our result provides substantial progress as compared to @xcite .
here we collect some technical background which was used in section 2 above .
all of this is known .
we either provide references or , for convenience , in some cases sketch the proof .
proofs of the following improved combes - thomas bound can be found in @xcite ( where it was first observed ) and @xcite .
we state it here under assumptions which are sufficient for our applications . in particular , we assume @xmath11 , while the result holds in arbitrary dimension for a suitably modified class of potentials . as above , for an open @xmath53 we denote by @xmath502 the restriction of @xmath503 to @xmath9 with dirichlet boundary conditions .
let us sketch the proof by combining several well known facts . to this end , let @xmath522 denote the @xmath523-th schatten class , i.e. the set of all bounded operators @xmath316 such that @xmath524 .
as @xmath525 , by theorem b.9.3 of @xcite we have @xmath526 for each @xmath527 and @xmath528 .
the proof provided in @xcite shows that @xmath529 can be chosen uniform in @xmath530 and @xmath184 . in the sense of quadratic forms it holds that @xmath531 for each open @xmath53 , i.e.@xmath532 for all @xmath528 , see e.g. section vi.2 of @xcite . thus @xmath533
the hlder property of schatten classes implies @xmath534 uniformly in @xmath521 , @xmath184 and @xmath53 open . from the resolvent identity @xmath535
we easily see that @xmath536 holds uniformly also in the additional parameter @xmath537 . by proposition [ prop : ct ]
we also have @xmath538 and @xmath539 such that @xmath540 . as we may choose @xmath541 , follows from and by interpolation , more precisely from the fact that @xmath542 and @xmath543 .
recall that an operator @xmath316 is called dissipative if @xmath544 for all @xmath545 .
it is called maximally dissipative if it has no proper dissipative extension .
below we also use the notation @xmath546 for lebesgue measure in @xmath547 .
[ prop : weakbound ] there exists a universal constant @xmath47 such that for every separable hilbert space @xmath548 , every maximally dissipative operator @xmath316 in @xmath548 with strictly positive imaginary part ( i.e. @xmath549 for some @xmath44 and all @xmath545 ) , for arbitrary hilbert - schmidt operators @xmath550 , @xmath551 in @xmath548 , for arbitrary bounded non - negative operators @xmath552 , @xmath553 in @xmath548 , and for all @xmath554 the following holds : @xmath555 ^ 2 : & \:\|m_1 u_1^{1/2 } ( a - v_1u_1- v_2u_2)^{-1 } u_2^{1/2 } m_2\|_{{\textup{hs } } } > t \bigr\}\bigr\rvert\notag\\ & \leq c \|m_1\|_{{\textup{hs } } } \|m_2\|_{{\textup{hs } } } \cdot \frac{1}{t}.\end{aligned}\ ] ] this follows with layer - cake integration , which gives for the l.h.s . of @xmath559
^ 2:\ : { \lvert\ldots\rvert } > t^{1/s}\rbrace}\bigr\rvert\,{\mathrm{d}}t.\ ] ] the integrand is bounded by the minimum of @xmath560 and a bound following from . splitting the integral accordingly leads to .
\(1 ) the use of the interval @xmath561 $ ] as support of @xmath562 in proposition [ prop : weakbound ] and corollary [ cor : fracmoments ] is not essential . using shifting and scaling it can be replaced by an arbitrary compact interval @xmath563 , with constants becoming @xmath563-dependent .
s. n. , p. s. and g. s. gratefully acknowledge financial support by the university paris 7 denis diderot , where part of this work was done .
s. n. and g. s. also received support through us - nsf grant no .
dms-0245210 , p. s. received support from the dfg . a. boutet de monvel and a. surkova : localisation des tats de surface pour une classe doprateurs de schrdinger discrets potentiels de surface quasi - priodiques , _ helv .
phys.acta_ * 71*(5 ) , ( 1998 ) 459490 .
v. jaki , s. molchanov and l. pastur : on the propagation properties of surface waves , in `` wave propagation in complex media ( minneapolis , mn , 1994 ) , '' i m a math .
* 96 * , springer , new york , 1998 , pp . 143154 .
d. hundertmark and w. kirsch : spectral theory of sparse potentials , in `` stochastic processes , physics and geometry : new interplays , i ( leipzig , 1999 ) , '' amer . math .
providence , ri , 2000 , pp . | we present a new , short , self - contained proof of localization properties of multi - dimensional continuum random schrdinger operators in the fluctuation boundary regime .
our method is based on the recent extension of the fractional moment method to continuum models in @xcite , but does not require the random potential to satisfy a covering condition .
applications to random surface potentials and potentials with random displacements are included . |
since 1998 , the type ia supernova ( sne ia ) observations @xcite have shown that our universe has entered into a phase of accelerating expansion . during these years from that time , many additional observational results , including current cosmic microwave background ( cmb ) anisotropy measurement from wilkinson microwave anisotropy probe ( wmap)@xcite , and the data of the large scale structure ( lss ) from sloan digital sky survey ( sdss ) @xcite , also strongly support this suggestion .
these observational results have greatly inspirited theorists to understand the mechanism of the accelerating expansion of the universe , which is usually attributed to an exotic energy component with negative pressure , dubbed dark energy ( de ) .
the simplest but most natural candidate of de is the cosmological constant @xmath17 , with the constant equation of state ( eos ) @xmath18 .
as we know , the cosmic concordance model confronts with two difficulties : the fine - tuning problem and the cosmic coincidence problem .
both of these problems are related to the de density . in order to solve or alleviate cosmological constant puzzles ,
many dynamical de models are proposed , where the de density and its eos are time - varying .
however , the predictions of the cosmological constant model still fit to the current observations @xcite .
therefore the dynamical de models being proposed should not be far away from the cosmological constant model , such as quintessence @xcite , phantom @xcite , quintom @xcite , k - essence @xcite , tachyon @xcite , ghost condensate @xcite , holographic de @xcite and agegraphic de @xcite etc .
although many de models have been presented , the nature of de is still a conundrum .
this puzzle can not be understood before a complete theory of quantum gravity is established .
but the two additional aspects from the current cosmological observations and some basic quantum gravitational principles may shed light on probing the nature of de .
on the one hand , provided that we know little on the theoretical nature of de at present , the combined cosmic observations can play an important role in understanding the nature of de .
the cosmological parameters space in the de model can be determined by the constraints of the data combinations .
recently , the 397 sn ia data was compiled in ref .
@xcite by adding cfa3 sample from the cfa sn group to the union set by ref .
@xcite , which include 250 sn ia at high redshift but only 57 at low redshift , to form the constitution set .
aside from the sn ia data , the combined analysis is required in order to break the degeneracy between the cosmological parameters , which includes cosmic observations from baryon acoustic oscillations ( bao ) , cmb and the observational hubble data ( ohd ) .
the bao are detected in the clustering of the combined 2dfgrs and sdss main galaxy samples or the sdss luminous red galaxies and measure the distance - redshift relation . from these samples ,
the values of @xmath19 $ ] and their inverse covariance matrix in the measurement of bao can be obtained @xcite . for the measurement of cmb
, we utilize the shift parameter @xmath20 at the photon decoupling epoch @xmath21 , the acoustic scale @xmath22 , and together with the physical baryon density parameter multiplied by 100 , thus it is @xmath23 @xcite . here , it is worth noting that the wmap distance information @xmath24 and @xmath22 can not be measured by wmap directly , but are derived from making a global fitting constraint with mcmc method by using the full wmap data on the assumption that a certain cosmological model has been given in advance @xcite .
although in theory the inverse covariance matrix on @xmath24 and @xmath22 is model dependent , it is feasible to use the derived results about @xmath24 and @xmath22 to constrain the parameters in another de model since @xmath24 and @xmath22 do not depend strongly on the de model which is not far away from the cosmological constant model @xcite .
what is more , the paper @xcite has been demonstrated that @xmath25 $ ] effectively provide a good summary of cmb data when the de model parameters are constrained .
in addition , we employ the ohd at twelve different redshifts determined by using the differential ages of passively evolving galaxies in ref .
@xcite , where the value of the hubble constant is replaced by @xmath26 in ref .
@xcite , and add the three more observational data @xmath27 and @xmath28 in @xcite . since the constraint results of a given model are dependent on the combined data @xcite , in this paper we use a fully combined observations from the 397 sn ia standard candle data , the value of @xmath29 $ ] and their inverse covariance matrix in the measurement of bao , the values of @xmath25 $ ] and their inverse covariance matrix in the measurement of cmb , and the fifteen ohd .
on the other hand , the models which are constructed in light of some fundamental principle are more charming , since this kind of de model may exhibit some underlying features of de , for instance the holographic de model @xcite and the agegraphic de model @xcite .
the holographic de model is built on the basis of holographic principle and some features of quantum gravity theory .
the agegraphic de model is derived from taking the combination between the uncertainty relation in quantum mechanics and general relativity into account . in this paper
, we focus on the holographic de model , which is considered as a dynamic vacuum energy .
according to the holographic principle , the number of degrees of freedom in a bounded system should be finite and is related to the area of its boundary . by applying the principle to cosmology , one can obtain the upper bound of the entropy contained in the universe . for a system with size @xmath30 and uv cut - off @xmath17 without decaying into a black hole , it is required that the total energy in a region of size @xmath30 should not exceed the mass of a black hole of the same size , thus @xmath31 .
the largest @xmath30 allowed is the one saturating this inequality , thus we obtain the holographic de density @xmath32 where c is a numerical constant and @xmath33 is the reduced planck mass @xmath34 .
it just means a duality between uv cut - off and ir cut - off .
the uv cut - off is related to the vacuum energy , and ir cut - off is related to the large scale of the universe , for example hubble horizon , particle horizon , event horizon , ricci scalar or the generalized functions of dimensionless variables as discussed by @xcite .
next , we give a brief review on the main results when hubble horizon , particle horizon , event horizon or ricci scalar are taken as the ir cut - off , respectively .
@xmath35 @xmath36 .
as pointed in @xcite , it is found that the holographic de density is in proportion to @xmath37 , the same as dark matter density , i.e. @xmath38 .
it appears that it is natural to solve the coincidence problem .
however , hsu @xcite pointed out that the dark energy eos @xmath39 was obtained in this instance .
it is obvious that this result is not consistent with the current observations .
this bad situation can be changed by considering the holographic de with hubble horizon as the time variable cosmological constant .
more detailed analysis is presented in ref .
@xcite .
@xmath35 @xmath40 .
as shown in paper @xcite , li pointed out that this yields the dark energy eos is not less than @xmath41 .
thus the current accelerated expansion of our universe can not be well explained .
however , this result in @xcite is obtained on the assumption that de dominates .
the holographic de model with particle horizon has been discussed in detail by @xcite .
@xmath35 @xmath42 . the holographic de model with event horizon
can reveal the dynamic nature of the vacuum energy and provide a desired eos of the holographic de with the model parameter @xmath43 .
furthermore , the holographic de behaves like quintessence , cosmological constant and phantom respectively for the different values of the model parameter : @xmath44 , @xmath45 and @xmath46 @xcite .
therefore , the value of model parameter @xmath43 plays a crucial role in determining the property of holographic de in this case . however , this model is confronted with the causality problem : why should the present density of de be determined by the future event horizon of the universe . @xmath35 @xmath47 . in @xcite , it has shown that this model can avoid the causality problem and naturally solve the coincidence problem of dark energy after ricci scalar is taken as the ir cut - off and the parameters have been well constrained by the combined astronomical observations @xcite . subsequently , in @xcite , granda and oliveros generalized the form of the ir cut - off on the basis of the ricci scalar : @xmath48 where there are two independent model parameters @xmath49 and @xmath50 , which can be determined by using the combined constraints of the thorough observational datasets . in this paper , we consider the holographic de model with new ir cut - off in both flat and non - flat case .
the performance of a global fitting will be made by using the markov chain monte carlo ( mcmc ) method . in this way
, we can work in the framework of multi - parameter freedoms , including the basic cosmological parameters ( @xmath51 ) and the new - added model parameters ( @xmath52 ) .
the paper is organized as follows . in next section ,
we briefly review the holographic de model with new ir cut - off . in section iii
, we perform the cosmic observation constraint on the holographic de model .
the last section is the conclusion .
in this section , we give a brief review on the general formula in the holographic de model with new ir cut - off . with a friedmann - robertson - walker ( frw ) metric @xmath53,\end{aligned}\ ] ] the einstein field equation can be written as @xmath54 where @xmath55 is the hubble function , and @xmath56 and @xmath57 are the energy density and the pressure of a general piece of matter , and their subscripts @xmath58 denote @xmath59 , @xmath60 and @xmath61 , which respectively correspond to matter component , the holographic de with new infrared cut - off and the curvature part of space . here
the matter component includes the cold dark matter and the baryon matter , i.e. @xmath62 the parameter @xmath63 denote the closed , flat and open geometries , respectively . the effective energy density and the effective pressure of the curvature part are @xmath64 as suggested by granda and oliveros in paper @xcite , the energy density of the holographic de with new ir cut - off is given as @xmath65 where @xmath49 and @xmath50 are the dimensionless parameters in holographic de model with new ir cut - off , which are regarded as independent of each other . in this paper ,
a dot denotes a derivative with respect to the cosmic time @xmath66 . after changing the variable from the cosmic time @xmath66 to @xmath67
, we can rewritten the eq .
( 4 ) as @xmath68 with the help of the definitions as follows : @xmath69 the eq .
( 10 ) can be ulteriorly rewritten as @xmath70 solving this first order differential equation about @xmath71 , we can obtain @xmath72 where @xmath73 is the integral constant and can be derived from the initial condition @xmath74 , which is @xmath75 , and @xmath76 is the dimensionless energy density of the holographic de with new ir cut - off : @xmath77 then , combining the above definition of the dimensionless energy density of the holographic de with its conservation equation , we can obtain the eos of the holographic de with new ir cut - off @xmath78 in addition , we shall investigate the evolution of the deceleration parameter .
combining eqs .
( 4 ) and ( 5 ) with the definition of the deceleration parameter , we can get @xmath79 where we have utilized the definitions of @xmath80 and @xmath81 . according to the energy conservation equation , we have @xmath82 in the review above , it is direct and natural to consider the parameters @xmath83 and @xmath84 when we solve the differential eq .
next , we discuss three special cases when the denominators in eq .
( 13 ) equal zero as follows : * case 1 : * @xmath85 and @xmath86 in this case , we obtain @xmath87 and @xmath88 . now the energy density of the holographic de is @xmath89 .
compared to the friedmann equation , it is found that the de density is not consistent with the current observations of @xmath90 exotic component in the universe .
* case 2 : * @xmath91 and @xmath86 from the latter equation , we can get @xmath92 .
then the energy density of the holographic de is @xmath93 . in this case , the solution of the first order differential equation is @xmath94 where we have set the integral constant @xmath95 and defined @xmath96 as the dimensionless energy density of the holographic de . considering the current value of the dimensionless de density ,
we can obtain the value of the only model parameter @xmath97 .
so the case is a viable de model .
* case 3 : * @xmath85 and @xmath98 from the former equation , we can obtain @xmath99 .
then the energy density of the holographic de is @xmath100 . in this case , the solution of the first order differential equation is @xmath101 where we have taken the integral constant @xmath102 and defined @xmath103 as the dimensionless energy density of the holographic de .
considering the current value of the dimensionless de density , we can get the value of the only model parameter @xmath104 .
so the case with the non - flat background geometry is a viable dark energy model .
the result in the flat case is the same as * case 1*. as far as the * case 2 * and * case 3 * discussed above are concerned , the model parameters @xmath49 and @xmath50 are reliant on each other . in this paper , we consider the generalized case with independent model parameters .
in the next section , according to the combined observational data , we restrict the basic cosmological parameters and the model parameters , all of which are independent on each other .
.the data fitting results of the cosmological parameters and the model parameters with @xmath105 , @xmath106 regions in the flat and non - flat holographic de model with new infrared cut - off and @xmath107 model ( flat case and non - flat case ) , where the combined observational data from sn @xmath108 , bao and cmb and ohd are used . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] the best fit values of the model parameters from observational hubble data @xcite are determined by minimizing @xmath109 ^ 2}{\sigma^2(z_i)},\label{eq : chi2h}\ ] ] where @xmath110 denotes the parameters contained in the model , @xmath111 is the predicted value for the hubble parameter , @xmath112 is the observed value , @xmath113 is the standard deviation measurement uncertainty , and the summation is over the @xmath114 observational hubble data points at redshifts @xmath115 . c. armendariz - picon , v. f. mukhanov and p. j. steinhardt , phys .
lett . * 85 * 4438 ( 2000 ) [ astro - ph/0004134 ] ; c. armendariz - picon , v. f. mukhanov and p. j. steinhardt , phys .
d * 63 * 103510 ( 2001 ) [ astro - ph/0006373 ]
. e. garcia - berro , e. gaztanaga , j. isern , o. benvenuto and l. althaus , arxiv : astro - ph/9907440 ; a. riazuelo and j. uzan , phys .
d * 66 * 023525 ( 2002 ) ; v. acquaviva and l. verde , jcap * 0712 * 001 ( 2007 ) .
s. nesseris and l. perivolaropoulos , phys .
d * 72 * 123519 ( 2005 ) ; l. perivolaropoulos , phys .
d * 71 * 063503 ( 2005 ) ; e. di pietro and j. f. claeskens , mon . not .
. soc . * 341 * 1299 ( 2003 ) ; a. c. c. guimaraes , j. v. cunha and j. a. s. lima , jcap * 0910 * 010 ( 2009 ) . | in this paper , the holographic dark energy model with new infrared ( ir ) cut - off for both the flat case and the non - flat case are confronted with the combined constraints of current cosmological observations : type ia supernovae , baryon acoustic oscillations , current cosmic microwave background , and the observational hubble data . by utilizing the markov chain monte carlo ( mcmc ) method
, we obtain the best fit values of the parameters with @xmath0 errors in the flat model : @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , @xmath6 . in the non - flat model ,
the constraint results are found in @xmath0 regions : @xmath7 , @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 , @xmath13 . in the best fit holographic dark energy models ,
the equation of state of dark energy and the deceleration parameter at present are characterized by @xmath14 ( flat case ) and @xmath15 ( non - flat case ) .
compared to the @xmath16 model , it is found the current combined datasets do not favor the holographic dark energy model over the @xmath16 model . |
for a given real @xmath0 let @xmath1 be the set of all @xmath2$]-valued random variables whose mean is equal to @xmath3 . formally
@xmath4= p \}.\ ] ] the main motivation behind this work is the following , well - known , problem .
+ [ hoeff ] fix @xmath5 real numbers @xmath6 and a real number , @xmath7 , such that @xmath8 . find ( or give upper bounds on ) @xmath9,\ ] ] where the supremum is taken over all random vectors @xmath10 of independent random variables with @xmath11 , for @xmath12 .
if @xmath13 , then the problem is trivial ; just choose @xmath14 to be equal to @xmath15 with probability @xmath16 .
there is a vast amount of literature that is related to problem [ hoeff ] .
the interested reader is invited to take a look at the works of bentkus @xcite , @xcite , @xcite , fan et al .
@xcite , from @xcite , from et al .
@xcite , gyrfi et al .
@xcite , hoeffding @xcite , kha et al .
@xcite , krafft et al .
@xcite , mcdiarmid @xcite , pinelis @xcite,@xcite , schmidt et al .
@xcite , siegel @xcite , talagrand @xcite , xia @xcite among many others . + determining the function @xmath17 , for given @xmath18 , turns out to be a notorious problem that has been around for many years . to our knowledge ,
no solution to this problem has ever been reported and most of the existing work focuses towards obtaining upper bounds on the function @xmath17 that are as tight as possible . +
probably the first systematic approach that allows one to obtain upper bounds on large deviations from the expectation for sums of independent , bounded random variables was performed by hoeffding in @xcite .
hoeffding s approach is based on a method of bernstein ( see @xcite , page @xmath19 ) and from now on will be referred to as the _ bernstein - hoeffding _ method .
the bernstein - hoeffding method is , briefly , the following .
+ markov s inequality and the assumption that the random variables are independent imply that @xmath20 \leq e^{-ht } \prod_{i=1}^{n } \mathbb{e}[e^{hx_i}]\leq e^{-ht}\left\{\frac{1}{n } \sum_{i=1}^{n}\mathbb{e}[e^{hx_i } ] \right\}^n , \ ; \text{for all}\ ; h>0,\ ] ] where the last inequality comes from the arithmetic - geometric means inequality . by exploiting the fact that the function @xmath21 is _ convex
_ one can show that @xmath22\leq \mathbb{e}[e^{hb_i}],\ ] ] where @xmath23 is a bernoulli random variable of mean @xmath15 .
hence we conclude that @xmath20 \leq e^{-ht } \{(1-p)+ pe^h\}^n , \ ; \text{for all}\ ; h>0,\ ] ] where @xmath24 .
if we minimise the expression in the right hand side of the last inequality with respect to @xmath25 , we find @xmath26 and hence we obtain the following celebrated result of hoeffding ( see @xcite , theorem @xmath16 ) .
+ [ folklore ] let @xmath27 be @xmath5 , given , real numbers from the interval @xmath28 .
let also the random variables @xmath29 be independent and such that @xmath30 , for each @xmath31
. set @xmath32 $ ] .
then for any @xmath7 such that @xmath33 we have @xmath34 \leq \inf_{h>0}\left\ { e^{-ht}\left(1-p + pe^h\right)^n\right\ } .\ ] ] furthermore , @xmath35 the function @xmath36 in the last expression is the so - called _ hoeffding bound _ ( or _ hoeffding function _ ) on tail probabilities for sums of independent , bounded random variables . throughout this paper
, we will denote by @xmath37 a bernoulli random variable with mean @xmath38 and by @xmath39 a binomial random variable of parameters @xmath5 and @xmath38 . if two random variables @xmath40 have the same distribution we will write @xmath41 .
we remark that the hoeffding bound is sharp , in the sense that the bernoulli random variables @xmath42 attain the bound , i.e. , @xmath43 \right\}^n = h(n , p , t ) , \ ] ] where @xmath23 is a @xmath42 random variable .
the main ideas behind this work are hidden in the fact that @xmath44 = \mathbb{e}\left[e^{hb}\right],\ ] ] where @xmath45 is a binomial random variable of parameters @xmath5 and @xmath46 $ ] , and the fact that the function @xmath47 , is non - negative , increasing and convex . in a subsequent section
we will show that , while applying the bernstein - hoeffding method , one can replace the exponential function @xmath48 , with any function @xmath49 having the aforementioned properties .
let us mention that hoeffding considered the tail probability @xmath50 $ ] , where @xmath51 , instead of the tail @xmath52 $ ] , where @xmath53 , thus obtaining a bound that looks different from the bound of the previous theorem .
the reader is invited to verify that the above bound is the same as the bound given by formula @xmath54 in @xcite .
we choose to work with the tail @xmath52 $ ] because it fits better to our goals
. a slightly looser but more widely used version of hoeffding s bound is the function @xmath55 , which follows from the fact that @xmath56 ( see @xcite , formula ( 2.3 ) ) .
+ there exists quite some work dedicated to improving hoeffding s bound .
see for example the work of bentkus @xcite , pinelis @xcite , siegel @xcite and talagrand @xcite , just to name a few references .
let us bring the reader s to attention the following two results that are extracted from the papers of talagrand ( @xcite , theorem @xmath57 ) and bentkus ( @xcite , theorem @xmath57 ) .
talagrand s paper focuses on obtaining some `` missing '' factors in hoeffding s inequality whose existence is motivated by the central limit theorem ( see @xcite , section @xmath16 ) .
these factors are obtained by combining the bernstein - hoeffding method together with a technique ( i.e. suitable change of measure ) that is used in the proof of cramr s theorem on large deviations , yielding the following .
+ [ talagr ] let @xmath27 be @xmath5 , given , real numbers from the interval @xmath28 .
let also the random variables @xmath29 be independent and such that @xmath30 , for each @xmath31 .
set @xmath24 .
then , for some absolute constant , @xmath58 , and every real number @xmath7 such that @xmath59 , we have @xmath60\leq \left\ { \theta\left(\frac{t - np}{\sqrt{np(1-p)}}\right)+ \frac{k}{\sqrt{np(1-p ) } } \right\}\cdot h(n , p , t ) , \ ] ] where @xmath36 is the hoeffding bound and @xmath61 is a non - negative function such that @xmath62 see @xcite for a proof of this theorem and for a precise definition of the function @xmath61 .
in other words , talagrand s result improves upon hoeffding s by inserting a `` missing '' factor of order @xmath63 in the hoeffding bound . notice that talagrand s result holds true for @xmath64 $ ] , for some absolute constant @xmath58 whose value does _ not _ seem to be known .
talagrand ( see @xcite , page 692 ) mentions that one can obtain a rather small numerical value for @xmath58 , but numerical computations are left to others with the talent for it .
one of the purposes of this paper is to improve upon hoeffding s inequality by obtaining `` missing '' factors with exact numerical values for the constants .
+ part of bentkus paper performs comparisons between @xmath65 $ ] and tails of binomial and poisson random variables .
a crucial idea in the results of @xcite is to compare @xmath65 $ ] with means of particular functions of certain random variables . in particular , in the proof of theorem @xmath57 in @xcite
one can find the following result .
+ [ ben ] let the random variables @xmath29 be independent and such that @xmath66 , for each @xmath31
. set @xmath32 $ ] .
then , for any positive real , @xmath7 , such that @xmath33 , we have @xmath34 \leq \inf_{a < t } \frac{1}{t - a } \mathbb{e}\left[\max\{0 , b - a\ } \right ] , \ ] ] where @xmath45 .
furthermore , if @xmath7 is additionally assumed to be a _ positive integer _
, we have @xmath34 \leq e \cdot\mathbb{p}\left [ b \geq t\right],\ ] ] where @xmath67 .
the quantity on the right hand side of the first inequality is estimated in @xcite , lemma @xmath68 .
we will see in the forthcoming sections that first statement of bentkus result is optimal in a slightly broader sense , i.e. , it is the best bound that can be obtained from the inequality @xmath69
\leq \frac{1}{f(t)}\mathbb{e}[f(b ) ] , \ ] ] where @xmath70 is a non - negative , convex and increasing function . additionally
, we will improve upon the constant @xmath71 of the second statement .
in this paper we shall be interested in employing the bernstein - hoeffding method to a larger class of generalized moments .
such approaches have been already performed by bentkus @xcite , eaton @xcite , pinelis @xcite,@xcite . nevertheless , we were not able to find a systematic study of the classes of functions that are considered in our paper .
we now proceed by defining a class of functions that is appropriate for the berstein - hoeffding method .
let us call a function @xmath72 _ sub - multiplicative _ if @xmath73 , for all @xmath74 .
we will denote by @xmath75 the set of all functions @xmath72 that are sub - multiplicative , increasing and convex .
examples of such functions are @xmath76 , for fixed @xmath77 , @xmath78 , for fixed @xmath79 and so on .
our first result shows that the bernstein - hoeffding method can be adjusted to the class @xmath75 .
+ [ mainhoeff ] let @xmath75 be defined as above . let the random variables @xmath29 be independent and such that @xmath66 , for each @xmath31 . set @xmath32 $ ] .
then , for any positive real , @xmath7 , such that @xmath33 , we have @xmath34 \leq \inf_{f\in \mathcal{f}_{sic}}\left\ { \frac{1}{f(t)}\left\{(1-p)f(0 ) + pf(1)\right\}^n\right\ } .\ ] ] we prove this result in section [ mainhof ] .
theorem [ mainhoeff ] can be deduced using the same argument as hoeffding s result .
its proof ought to be somewhere in the literature but we were unable to locate a reference .
we provide a proof for the sake of completeness . additionally , we prove in section [ mainhof ] that hoeffding s bound is the best bound that can be obtained using functions from the class @xmath75 . in section [ improve ] we extend theorem [ mainhoeff ] to an even larger class of moments .
more precisely , fix @xmath80 and let @xmath81 be the set consisting of all _ convex _ functions @xmath82 that are _ increasing _ in the interval @xmath83 .
examples of such functions are @xmath84 for fixed @xmath85 , @xmath86 for fixed @xmath85 , @xmath76 for @xmath77 and so on . in section
[ improve ] , by employing ideas from the theory of _ convex orders _ , we obtain the following .
+ [ maintwo ] let the random variables @xmath29 be independent and such that @xmath66 , for each @xmath31
. set @xmath32 $ ] .
then , for any fixed real number , @xmath7 , such that @xmath33 , we have @xmath34 \leq \inf_{f\in \mathcal{f}_{ic}(t)}\frac{1}{f(t)}\mathbb{e } [ f(b ) ] , \ ] ] where @xmath45 is a binomial random variable and @xmath81 is the class of functions defined above . in section [ optopt ] we show that the functions @xmath87 that minimise @xmath88 $ ] are those used in the aforementioned result of bentkus , i.e. , theorem [ ben ] .
we then choose a particular function @xmath89 and obtain a version of talagrand s result having exact numerical constants .
more precisely , in section [ missingtwo ] , we prove the following improvement upon hoeffding s inequality .
+ [ talagrtwo ] let the random variables @xmath29 be independent and such that @xmath66 , for each @xmath31
. set @xmath32 $ ] .
let @xmath7 be a fixed _ positive integer _
such that @xmath90 .
then @xmath34 \leq \frac{1+h}{e^h}\cdot\left ( h(n , p , t)- t(n , p , t;h ) \right ) + \left(1-\frac{1+h}{e^h}\right ) \mathbb{p}\left[b_{n , p } = t\right],\ ] ] where @xmath36 is the hoeffding function , @xmath91 is a binomial random variable of parameters @xmath5 and @xmath3 , @xmath92 , \ ] ] and @xmath25 is such that @xmath26 , i.e. , it is the optimal real such that @xmath93 = \inf_{s>0}\ ; \frac{1}{e^{st } } \mathbb{e}[e^{sb}],\ ] ] with @xmath45 .
let us illustate that the bound of the previous result is an improvement upon hoeffding s inequality .
indeed , notice that the bound of the previous theorem is @xmath94\ ] ] and the later quantity is a convex combination of @xmath36 and @xmath95 $ ] . now hoeffding s theorem [ folklore ] implies that @xmath96\leq \mathbb{p}\left[b_{n , p } \geq t\right ] \leq h(n , p , t)\ ] ] and therefore the bound of the previous result is _ smaller _ than hoeffding s .
+ in brief , the previous result improves upon hoeffding s by adding a `` missing '' factor that is equal to @xmath97 .
since @xmath26 it follows that the `` missing '' factor can be written as @xmath98 on the other hand , talagrand s result provides a factor that is approximately @xmath99 is it unclear how to compare the two factors without knowing the constant @xmath58 . if we assume that @xmath100 then elementary , though quite tedious , calculations show that talagrand s bound is sharper than the bound of theorem [ talagrtwo ] .
our bound has the advantage that it does _ not _ involve unknown constants and that it is obtained using a rather simple argument .
+ using theorem [ maintwo ] we can also obtain the following , partial , improvement upon the second statement of bentkus result , i.e. , theorem [ ben ] .
+ [ binbin ] let the random variables @xmath29 be independent and such that @xmath66 , for each @xmath31 . set @xmath32 $ ] . then , for any fixed _ positive integer _ , @xmath7 , such that @xmath33
, we have @xmath34 \leq \frac{t - tp}{t - np}\cdot\mathbb{p } [ b\geq t ] , \ ] ] where @xmath45 is a binomial random variable .
note that for large @xmath7 , say @xmath101 , the previous result gives an estimate for which @xmath102 .
however , for values of @xmath7 that are close to @xmath103 , the previous result provides estimates for which @xmath104 can be arbitrarily large .
+ in section [ bernhoeff ] we generalise the bernstein - hoeffding method to sums of bounded , independent random variables for which the first @xmath105 moments are known .
more precisely , for given real numbers @xmath106 , let @xmath107 be the set of all @xmath2$]-valued random variables whose @xmath108-th moment equals @xmath109 .
formally , @xmath110=\mu_1,\mathbb{e}[x^2]=\mu_2\ldots , \mathbb{e}[x^m]=\mu_m\ } .\ ] ] notice that the set may be empty .
note also that , if @xmath107 is non - empty then we have @xmath111 .
recall the definition of the class @xmath75 , defined above .
the main result of section [ bernhoeff ] is the following .
+ [ moments ] fix positive integers , @xmath112 and for @xmath31 let @xmath113 be a sequence of real numbers such that @xmath114 and for which the class @xmath115 is non - empty .
let @xmath29 be independent random variables such that @xmath116 , for @xmath31 , and fix @xmath117 $ ]
. then @xmath60 \leq \inf_{f\in \mathcal{f}_{sic } } \;\frac{1}{f(t ) } \left\{\mathbb{e}\left [ f(t_{nm})\right]\right\}^n,\ ] ] where @xmath118 is the random variable that takes on values in the set @xmath119 and , for @xmath120 , it satisfies @xmath121 = \frac{1}{n}\sum_{i=1}^{n}\binom{m}{j } \mathbb{e}\left[x_i^j(1-x_i)^{m - j}\right ] .\ ] ] to our knowledge , this is the first result that considers the performance of the method under additional information on higher moments .
notice that the probability distribution of the random variable @xmath118 does _ not _ depend on the random variables @xmath29 . indeed , using the binomial formula ,
it is easy to see that @xmath122 = \sum_{k=0}^{m - j}\binom{m - j}{k}(-1)^{m - j - k } \mu_{i , m - k}\ ] ] and so @xmath118 is uniquely determined by the given sequences on moments @xmath123 .
we will refer to the random variable that takes values on the set @xmath124 with probability @xmath125 $ ] , for @xmath116 , as a @xmath115-_bernstein random variable_. let us also mention that bernstein random variables occur in the study of the so - called _ hausdorff moment problem _ ( see feller @xcite ) .
a similar result holds true for the class @xmath81 ; we state this result in section [ bernhoeff ] and sketch its proof . in section [ refinebinbin ]
we perform comparisons between @xmath65 $ ] and binomial tails that depend on the additional information on the moments . in section [ bernhoeffgame ]
we study the performance of the method on a certain class of bounded random variables that contain additional information on conditional means and/or conditional distributions .
we find random variables that are larger , in the sense of convex order , than any random variable from this class and prove similar results as above that take into account the additional information .
our approach is based on the notion of mixtures of random variables .
additionally , we construct random variables @xmath126 that are _ different _ from bernstein random variables and are larger , in the sense of convex order , than any random variable from the class @xmath127 , consisting of all random variables in @xmath1 whose variance is @xmath128 . in particular , in section [ bernhoeffgame ]
we prove the following .
+ [ xitheorem ] fix positive integer @xmath5 and assume that , for @xmath31 , we are given a pair @xmath129 for which the class @xmath130 is non - empty .
let @xmath29 be independent random variables such that @xmath131 , for @xmath31 .
set @xmath132 $ ] and fix @xmath7 such that @xmath53 .
then @xmath133 \leq \inf_{f\in \mathcal{f}_{ic}(t ) } \ ; \frac{1}{f(t ) } \mathbb{e}\left[f\left(\sum_{i=1}^{n}\xi_{p_i,\sigma_i } \right)\right ] , \ ] ] where , for @xmath31 the random variable @xmath134 is given by lemma [ xirandom ] .
furthermore , the infimum on the right hand side is attained by a function of the form @xmath135 , for some @xmath136 .
finally , in section [ unbounded ] we show that the bernstein - hoeffding method reduces to markov s inequality when employed to non - negative and _ unbounded _ random variables .
this subsection is devoted to the proof of theorem [ mainhoeff ] .
the proof will make use of the following elementary lemma , that is interesting on its own . + [
coupling ] fix real numbers @xmath137 such that @xmath138 .
let @xmath139 be a random variable that takes values on the interval @xmath140 $ ] and is such that @xmath141=p$ ] .
let @xmath142 be the random variable that takes on the values @xmath143 and @xmath144 with probabilities @xmath145}{b - a}$ ] and @xmath146-a}{b - a}$ ] , respectively .
then for any convex function , @xmath147\rightarrow \mathbb{r}$ ] , we have @xmath148 \leq \mathbb{e}[f(b ) ] .\ ] ] given @xmath139 , we couple the random variables by setting @xmath149 to be either equal to @xmath143 with probability @xmath150 , or equal to @xmath144 with probability @xmath151 . it is easy to see that @xmath152=x$ ] and so @xmath153= \mathbb{e}[\mathbb{e}[b_x|x ] ] = \mathbb{e}[x]=p .\ ] ] jensen s inequality now implies that @xmath148 = \mathbb{e}[f(\mathbb{e}[b_x|x ] ) ] \leq \mathbb{e}[f(b_x|x)]=\mathbb{e}[f(b_x ) ] , \ ] ] as required .
we are now ready to prove our first main result .
set @xmath154 and fix a function @xmath155 . by markov s inequality , independence and the assumption that @xmath70 is increasing and submultiplicative , we conclude that @xmath156 & \leq&\mathbb{p}[f(s_n ) \geq f(t)]\quad ( f \;\text{is increasing})\\ & \leq & \frac{1}{f(t ) } \mathbb{e}\left[f\left(\sum_{i=1}^{n}x_i\right ) \right]\ ; ( \text{markov 's inequality } ) \\ & \leq & \frac{1}{f(t ) } \mathbb{e}\left[\prod_{i=1}^{n } f(x_i ) \right ] \ ; ( \text{sub - multiplicativity } ) \\ & \leq & \frac{1}{f(t ) } \prod_{i=1}^{n } \mathbb{e}[f(x_i ) ] \ ; ( \text{independence } ) .\end{aligned}\ ] ] since the function @xmath157 is convex , lemma [ coupling ] implies that @xmath158 \leq \mathbb{e}[f(b_i ) ] , \ ; \text{where } \ ;
b_i \sim \text{ber}(p_i ) .\ ] ] hence @xmath159\leq \frac{1}{f(t)}\prod_{i=1}^{n } \mathbb{e}[f(b_i ) ] .\ ] ] now the arithmetic - geometric means inequality yields @xmath160 \leq \left\ { \frac{1}{n } \sum_{i=1}^{n } \mathbb{e}[f(b_i ) ] \right\}^n = \left\ { ( 1-p)f(0)+pf(1)\right\}^n\ ] ] and thus @xmath159\leq \frac{1}{f(t ) } \left\{(1-p)f(0)+pf(1)\right\}^n , \ ; f\in \mathcal{f}_{sic } .\ ] ] the result follows . the first statement in hoeffding s result ( theorem [ folklore ] )
is obtained by adjusting the previous proof to the function @xmath161 , which clearly belongs to @xmath75 . for @xmath155 , let @xmath162 notice that theorem [ mainhoeff ] suggests that there may be some space for improvements upon hoeffding s bound , i.e. , there may exist a function @xmath163 such that @xmath164 for all @xmath25 .
we now show that this is not the case , when @xmath7 is an _
integer_. the following result solves the problem of finding @xmath165 in case @xmath7 is a positive _ integer_. + [ integer ] let @xmath7 be a positive integer .
suppose that @xmath166 is such that @xmath167 for all @xmath168 .
then there exists @xmath169 such that @xmath170 and @xmath171 , for some positive constant @xmath25 . since @xmath49 is sub - multiplicative and non - negative , it easy to see that @xmath172 . for @xmath173 ,
set @xmath174 .
then @xmath175 , @xmath176 and so @xmath177 since @xmath7 is a positive integer , it follows that @xmath178 .
hence @xmath179 .
the result follows upon setting @xmath180 . hence , for integer @xmath7 , we have @xmath181 , where the infimum is taken over all functions @xmath155 and @xmath36 is the hoeffding function that is defined in the introduction .
quoting hoeffding ( see @xcite , page @xmath182 ) , the bound @xmath20 \leq h(n , p , t)\ ] ] is the best that can be obtained from the inequality @xmath20 \leq e^{-ht } \prod_{i=1}^{n } \mathbb{e}[e^{hb_i}],\ ; h>0,\ ] ] where @xmath183)$ ] .
this follows from the fact that @xmath36 is obtained by minimising the expression on the right hand side of the above inequality with respect to @xmath77 .
proposition [ integer ] shows that , in case @xmath7 is a positive integer , hoeffding s bound is the best the can be obtained in a slightly broader sense , i.e. , @xmath36 is the best bound on @xmath52 $ ] that can be obtained by minimising @xmath184 $ ] with respect to @xmath185 .
in this section we prove theorem [ maintwo ] and show that the hoeffding bound can be improved using a larger class of functions , namely the class @xmath81 , defined in the introduction . once again , theorem [ maintwo ] implies that there may be some space for improvement upon hoeffding s bound .
we will employ this result and en route find a function @xmath89 such that @xmath186 < \inf_{h>0 } e^{-ht } \mathbb{e}[e^{hb } ] , \ ] ] where @xmath45 .
hence there is indeed space for improvement upon hoeffding s bound .
the proof of theorem [ maintwo ] will require some well - known results and the following notion of ordering between random variables ( see @xcite ) .
+ let @xmath139 and @xmath187 be two random variables such that @xmath148\leq \mathbb{e}[f(y ) ] , \ ; \text{for all convex functions } \ ; f:\mathbb{r}\rightarrow \mathbb{r},\ ] ] provided the expectations exist . then @xmath139 is said to be smaller than @xmath187 in the _ convex order _
, denoted @xmath188 .
the following two lemmas are well - known ( see theorems @xmath189 and @xmath190 in @xcite and theorem @xmath191 in @xcite ) .
the first one shows that convex order is closed under convolutions .
+ [ convexone ] let @xmath192 be a set of independent random variables and let @xmath193 be another set of independent random variables .
if @xmath194 , for @xmath31 , then @xmath195 the second lemma shows that a sum of independent bernoulli random variables is dominated , in the sense of convex order , by a certain binomial random variable .
+ [ convextwo ] fix @xmath5 real numbers @xmath27 from @xmath28 .
let @xmath196 be independent bernoulli random variables with @xmath197 .
then @xmath198 where @xmath199 is a binomial random variable of parameters @xmath5 and @xmath200 .
the proof of theorem [ maintwo ] is basically an extension of the proof of theorem [ mainhoeff ] .
fix @xmath87 .
since @xmath49 is non - negative and increasing in @xmath83 , markov s inequality implies that @xmath201 \leq \frac{1}{f(t ) } \mathbb{e}\left [ f\left(\sum_{i=1}^{n}x_i\right ) \right].\ ] ] since @xmath49 is convex , lemmata [ coupling ] and [ convexone ] imply that @xmath202 \leq \mathbb{e}\left [ f\left(\sum_{i=1}^{n}b_i\right ) \right ] , \ ] ] where @xmath203 ) , i=1,\ldots , n$ ] .
now lemma [ convextwo ] yields @xmath204 \leq \mathbb{e}\left [ f\left(b\right ) \right]\ ] ] and the result follows .
similar ideas as above have been employed to sums of independent bernoulli random variables by len and perron in @xcite . in a subsequent section
we employ theorem [ maintwo ] and en route improve upon hoeffding s inequality by inserting certain `` missing '' factors . before doing so
, we state some results regarding the optimal function in the class @xmath81 .
let the random variables @xmath29 be independent and such that @xmath66 , for each @xmath31 .
set @xmath32 $ ] and fix a real number , @xmath7 , such that @xmath33 .
we have shown in the previous section that , for @xmath87 , we have @xmath201 \leq \frac{1}{f(t)}\mathbb{e}\left [ f\left(b\right ) \right ] , \ ] ] where @xmath45 .
set @xmath205 = \frac{1}{f(t ) } \sum_{i=0}^{n } f(i)\cdot \mathbb{p}[b = i ] .\ ] ] in this section we solve the problem of finding @xmath206 , where the infimum is taken over all functions @xmath87 .
we show that the solution is related to bentkus result .
we begin with an observation on the optimal function . + [ opt ] let @xmath89 be a function such that @xmath207 , where the infimum is taken over all functions @xmath87 .
then we may assume that @xmath208 . if @xmath209 , then we set @xmath210 .
using this result we can find functions @xmath87 that minimise @xmath211 .
+ [ yuyipr ] let @xmath212 be such that @xmath207 , where the infimum is taken over all functions @xmath87 .
then @xmath213 equals @xmath214 , for some @xmath136 .
we may assume that @xmath208 and so @xmath215 is such that @xmath216 = t_n(\phi , t),\ ] ] where the infimum is taken over the set @xmath217 , containing all functions @xmath87 such that @xmath218 .
let @xmath219 be the smallest positive integer that is larger than @xmath7 .
note that , by definition , @xmath220 . for @xmath173 , define the function @xmath221 in other words , @xmath222 equals zero for @xmath223 and for @xmath224 it is a straight line starting from point @xmath225 and passing through the points @xmath226 and @xmath227 .
note that @xmath228 and that @xmath229 ; indeed , if @xmath230 , then @xmath231 and the function @xmath232 would be such that @xmath233\leq \mathbb{e}[\phi(b)]$ ] , which implies that @xmath234 $ ] is even worse than the bound obtained by markov s inequality , hence contradicts its optimality .
since the function @xmath235 is convex it follows that for every integer @xmath236 in the interval @xmath237 $ ] we have @xmath238 and this , in turn , implies that @xmath239 as required .
this yields the following .
+ [ yuyiproof ] let the parameters @xmath240 be as in theorem [ maintwo ] . then for any @xmath241 we have @xmath242 = \inf_{a < t } \;\frac{1}{t - a } \mathbb{e}\left[\max\{0 , b - a\ } \right ] , \ ] ] where @xmath199 and the infimum on the left hand side is taken over all functions @xmath87 . notice that we can write the function @xmath243 , for @xmath136 , in the form @xmath244 , where @xmath245 , and that this correspondence is injective . notice also that , since @xmath229 , we have @xmath246 .
the following question arises naturally from corollary [ yuyiproof ] .
+ what is the optimal @xmath247 such that @xmath248 = \mathbb{e}\left[\rho_{\varepsilon}(b ) \right ] \ ; ? \ ] ] we remark that such an @xmath247 will satisfy @xmath249 , where @xmath250 . to see this notice that if @xmath251 , then @xmath252 and we may decrease @xmath247 , until it reaches the point @xmath253 , without increasing the value @xmath254 $ ] . since @xmath249 it follows that @xmath255 .
now , finding the optimal @xmath247 is equivalent to finding the optimal @xmath25 .
we are _ not _ able to find this @xmath25 . nevertheless , due to the following result , one can easily find @xmath25 using , say , a binary search algotithm .
+ [ yuyiprtwo ] let the parameters @xmath240 be as in theorem [ maintwo ] . let @xmath77 be such that @xmath256 = \inf_{s>0 } \ ; \mathbb{e}\left[\max\{0 , s\cdot(b - t)+1\ } \right ] , \ ] ] where @xmath45 .
then we may assume that @xmath257 , for some positive integer @xmath258 .
recall that @xmath259 $ ] .
we have @xmath260 = \sum_{i=0}^{n}\binom{n}{i } p^i ( 1-p)^{n - i } \cdot g_h(i ) .\ ] ] the function @xmath261 $ ] is linear on the interval @xmath262 $ ] , for every @xmath263 .
hence the function @xmath264 is continuous and piecewise linear on the interval @xmath265 $ ] and this implies that it attains its minimum at the endpoints of @xmath262 $ ] , for some @xmath263 .
the result follows . in the next section
we obtain an improvement upon hoeffding s bound . in this section
we collect results that can be obtained by employing theorem [ maintwo ] .
we begin with the proof of theorem [ talagrtwo ] .
given @xmath77 define the function @xmath266 , for @xmath173 .
it is easy to see that @xmath87 .
let @xmath267 be the largest positive integer for which @xmath268 .
using theorem [ maintwo ] and the inequality @xmath269 , for @xmath270 , we estimate @xmath271 \leq \mathbb{e}[f(b)]&= & \sum_{i= m_t+1}^{n}(h(i
- t)+1)\mathbb{p}[b = i ] \\ &
< & \sum_{i = m_t+1}^{n}e^{h(i - t)}\mathbb{p}[b = i]\\ & \leq & h(n , p , t),\end{aligned}\ ] ] which shows that @xmath272 $ ] is strictly smaller than hoeffding s bound .
since we assume that @xmath273 it follows that @xmath274 which in turn implies , since @xmath7 is an integer , that @xmath275 , for all @xmath276 .
hence we can write @xmath277 & = & \sum_{i=0}^{n}e^{h(i - t)}\mathbb{p}[b = i ] - \sum_{i = t+1}^{n}(h(i - t)+1)\mathbb{p}[b = i ] \\ & = & \sum_{i=0}^{t-1}e^{h(i - t)}\mathbb{p}[b = i ] \\ & + & \sum_{i = t+1}^{n } \left(e^{h(i - t)}-(h(i - t ) + 1 ) \right)\mathbb{p}[b = i].\end{aligned}\ ] ] for @xmath278 , we have @xmath279 which implies that @xmath280 & \geq & \left ( 1- \frac{1+h}{e^h } \right ) h(n , p , t ) + \frac{1+h}{e^h}\cdot \sum_{i=0}^{t-1}e^{h(i - t)}\mathbb{p}[b = i]\\ & - & \left ( 1- \frac{1+h}{e^h } \right)\mathbb{p}\left[b_{n , p}=t\right ] . \end{aligned}\ ] ] the result follows .
if @xmath7 is _ not _ an integer , then one may use the previous bound with @xmath7 replaced by @xmath281 since @xmath60 \leq \mathbb{p}\left[\sum_{i=1}^{n}x_i\geq \lfloor t\rfloor \right ] .\ ] ] this result improves upon hoeffding s bound by fitting a `` missing '' factor that is equal to @xmath97 .
theorem [ maintwo ] allows to perform comparisons with binomial tails .
let @xmath282 so that @xmath283 and @xmath284 .
theorem [ maintwo ] implies that @xmath34 \leq \mathbb{e}[\psi(b)],\ ] ] where @xmath45 .
since @xmath7 is a positive integer , we can write @xmath285 = \sum_{i = t}^{n } ( i - t+1 ) \mathbb{p}[b = i ] = \sum_{i = t}^{n}\mathbb{p}[b\geq i ] .\
] ] now we use the following , well - known , estimate on binomial tails ( see feller @xcite , page 151 , formula @xmath286 ) : @xmath287 \leq \frac{i - ip}{i - np } \cdot\mathbb{p}[b = i ] , \ ; \text{for } \ ; i > np .\ ] ] therefore , @xmath285
\leq \sum_{i = t}^{n } \frac{i - ip}{i - np } \cdot\mathbb{p}[b = i ] \leq \frac{t - tp}{t - np } \cdot\mathbb{p}[b\geq t],\ ] ] as required . compare this result with the second statement of bentkus theorem [ ben ] , from section [ motrel ] .
note that for large @xmath7 , say @xmath101 , the previous result gives an estimate for which @xmath102 .
in a subsequent section we will show an extension of this result .
we begin this section with the proof of theorem [ moments ] .
the proof borrows ideas from the theory of bernstein polynomials ( see phillips @xcite , chapter @xmath288 ) . recall that , for a function @xmath289\rightarrow \mathbb{r}$ ] , the _ bernstein polynomial _ corresponding to @xmath70 is defined as @xmath290 for each positive integer @xmath105 .
the following is a folklore result regarding bernstein polynomials .
+ [ bernst ] if @xmath289\rightarrow [ 0,\infty)$ ] is convex , then @xmath291.\ ] ] if @xmath289\rightarrow [ 0,\infty)$ ] is continuous , then @xmath292 } |f(x)-b_{m}(f , x)| \rightarrow 0 , \ ; \text{as}\ ; m\rightarrow \infty .\ ] ] see @xcite theorems @xmath293 and @xmath294 .
we remark that the first statement is easy to prove and the second arose from bernstein s search for a proof of weierstrass theorem .
let @xmath155 .
since @xmath70 is non - negative , increasing and sub - multiplicative , markov s inequality implies that @xmath271 & \leq & \frac{1}{f(t ) } \mathbb{e}\left[f\left(\sum_{i=1}^{n}x_i\right)\right]\\ & \leq & \frac{1}{f(t ) } \prod_{i=1}^{n } \mathbb{e}\left[f\left(x_i\right)\right ] \\ & \leq & \frac{1}{f(t ) } \left\{\frac{1}{n}\sum_{i=1}^{n } \mathbb{e}[f(x_i ) ] \right\}^n , \end{aligned}\ ] ] where the last estimate comes from the arithmetic - geometric means inequality .
since @xmath70 is convex and @xmath295 $ ] , lemma [ bernst ] implies that @xmath296 \leq \mathbb{e}\left[b_m\left(f , x_i\right ) \right ] .\ ] ] now note that @xmath297 = \sum_{j=0}^{m } \binom{m}{j } \cdot\mathbb{e}\left[x_i^j(1-x_i)^{m - j } \right ] \cdot f(j / m).\ ] ] for @xmath298 let @xmath299 .\ ] ] notice also that @xmath300 = \sum_{k=0}^{m - j}\binom{m - j}{k}(-1)^{m - j - k}\mu_{i , m - k}\ ] ] which implies that @xmath301 $ ] is the same for all random variables from the clsss @xmath302 .
it is easy to verify that @xmath303 ; hence @xmath304 is a probability distribution on @xmath305 .
now , if we define the random variable @xmath118 that takes on the value @xmath306 with probability @xmath304 , we have @xmath307 \right\}^n = \frac{1}{f(t ) } \left\{\mathbb{e}\left [ f(t_{nm})\right]\right\}^n , \ ] ] as required . note that for @xmath308 the previous result reduces to theorem [ mainhoeff ] , from section [ mainhof ] .
in particular , we conclude the following generalisation of hoeffding s result .
+ with the same assumptions as in theorem [ moments ] , we have @xmath60 \leq \inf_{h>0 } e^{-ht } \sum_{j=0}^{m } \pi_j e^{h \frac{j}{m } } , \ ] ] where @xmath309 $ ] .
since @xmath310 converges uniformly to @xmath49 , as @xmath311 , we conclude that @xmath312 $ ] can be arbitrarily close to @xmath313 $ ] , provided that @xmath105 is sufficiently large .
recall the definition of the class @xmath81 from the introduction .
+ [ momopt ] fix positive integers , @xmath112 and for @xmath31 let @xmath113 be a sequence of reals such that @xmath114 and for which the class @xmath115 is non - empty .
let @xmath29 be independent random variables such that @xmath116 , for @xmath31 , and fix @xmath117 $ ] . then @xmath60 \leq \inf_{f\in \mathcal{f}_{ic}(t ) } \;\frac{1}{f(t ) } \mathbb{e}\left [ f(z_{nm})\right],\ ] ] where @xmath314 is an independent sum of random variables @xmath315 such that @xmath316 = \binom{m}{j } \cdot \mathbb{e}\left[x_i^j ( 1-x_i)^{m - j}\right ] , \;\text{for}\ ; j=0,1,\ldots , m .\ ] ] moreover , @xmath317 = \inf_{a < t } \ ; \frac{1}{t - a}\mathbb{e}[\max\{0,z_{nm}-a\}].\ ] ] the argument proceeds along the same lines as the proofs of the results in section [ improve ] and section [ optopt ] and so we only sketch it .
part of the proof of theorem [ moments ] yields @xmath318 , i.e. , @xmath319\leq \mathbb{e}[b_m(f , x_i)]=\mathbb{e}[f(z_i)]$ ] , for convex @xmath70 .
since the convex order is closed under convolutions , the first statement follows .
the proof of the second statement is almost identical to the proof of theorem [ yuyipr ] . in the previous result we found a random variable @xmath315 such that @xmath320 , for every @xmath116 .
note that @xmath321=\mathbb{e}[x_i]$ ] , for all @xmath31 .
however , higher moments of @xmath322 are not equal to the higher moments of @xmath14 .
let us illustrate this by assuming from now on that @xmath308 .
then @xmath323= \frac{1}{2}\mathbb{e}[x_i]+\frac{1}{2}\mathbb{e}[x_i^2 ] \geq \mathbb{e}[x_i^2]$ ] and so @xmath315 may _ not _ belong to @xmath324 .
notice that this is not the case when @xmath325 ; i.e. , when we consider random variables @xmath326 . in this case
( see theorem [ maintwo ] ) we were able to find bernoulli random variables @xmath23 from the class @xmath327 such that @xmath328\leq \mathbb{e}\left[f(b_i)\right]$ ] , for all functions @xmath87 .
the following question arises naturally from the above .
+ fix @xmath329 such that the class @xmath330 is non - empty .
does there exist random variable @xmath331 such that @xmath332\leq \mathbb{e}\left[f(\xi)\right]$ ] , for all @xmath333 and all increasing and convex functions @xmath82 ?
it turns out that the answer to the question is _ no_. in order to convince the reader we will use lemma [ cohhen ] below , taken from cohen et al .
let us first fix some notation .
if @xmath333 , let @xmath334 be its variance . set @xmath335 and and let @xmath336 be the random variable that takes on the values @xmath337 and @xmath16 with probability @xmath338 and @xmath339 , respectively .
it is easy to verify that @xmath336 has mean @xmath340 and variance @xmath128 .
the following result is proven in cohen et al .
@xcite and implies that @xmath336 has the maximum moments of any order , among all random variables in @xmath330 .
+ [ cohhen ] let @xmath333 and let @xmath336 be the random variable defined above
. then @xmath341 \leq \mathbb{e}\left[c^{k}\right]$ ] , for every non - negative integer @xmath236 .
see @xcite , lemma @xmath342 .
the following is an immediate consequence of the previous lemma
. + let @xmath333 . if @xmath139 is not the random variable @xmath336 of lemma [ cohhen ] , then @xmath343 < \mathbb{e}\left[e^{hc}\right]$ ] , for any @xmath77 .
note that the inequality in the conclusion is strict .
the previous lemma implies that @xmath341 \leq \mathbb{e}\left[c^{k}\right]$ ] , for every non - negative integer @xmath236 .
since @xmath139 is not equal to @xmath336 , there is at least one @xmath344 such that @xmath345 \leq \mathbb{e}\left[c^{k_0}\right]$ ] ; this follows from the fact that the sequence of moments uniquely determines that random variable ( see feller @xcite , chapter vii.3 ) .
therefore , taylor expansion implies that @xmath343 < \mathbb{e}\left[e^{hc}\right]$ ] .
the following result implies that the previous question has a negative answer .
+ [ impossible ] let @xmath346 be such that @xmath347 and set @xmath348 .
there does _ not _ exist random variable @xmath331 such that @xmath349 \leq \mathbb{e}\left[f(\xi)\right],\ ] ] for all @xmath333 and every @xmath87 .
we argue by contradiction .
suppose that such a @xmath350 does exist .
the previous corollary implies that @xmath350 must be the random variable @xmath336 , from lemma [ cohhen ] .
we now define a random variable @xmath351 as follows .
let @xmath352 be such that @xmath353= \frac{\sigma^2}{\mu_{1}^{2}+\sigma^2}\ ; \text{and}\ ; \mathbb{p}\left[c'=\frac{\mu_{1}^{2}+\sigma^2)}{\mu_1}\right]= \frac{\mu_{1}^{2}}{\mu_{1}^{2}+\sigma^2}.\ ] ] if @xmath335 is as in lemma [ cohhen ] , let us define the function @xmath354 , which is clearly increasing and convex .
it is easy to verify that @xmath355 = \frac{\sigma^2}{(1-\mu_1)^2+\sigma^2}\quad \text{and}\quad \mathbb{e}\left[g(c ' ) \right ] = \frac{\mu_1 \sigma^2}{(\mu_1 ^ 2+\sigma^2)((1-\mu_1)^2+\sigma^2)}.\ ] ] if we divide the last two equations we get @xmath356}{\mathbb{e}\left[g(c ' ) \right ] } = \frac{\mu_1 ^ 2 + \sigma^2}{\mu_1 } < 1,\ ] ] which contradicts the maximality of @xmath350 . in the next section
we exploit the fact that the random variable @xmath315 , from theorem [ momopt ] , is stochastically smaller than a particular binomial random variable . using this result
, we will obtain a refined version of theorem [ binbin ] .
we begin this section by recalling the following , well - known , result of hoeffding ( see @xcite , theorem @xmath191 ) .
+ [ lemhof ] fix a positive integer @xmath357 and let @xmath358 be real numbers from the interval @xmath28 .
let @xmath359 be independent bernoulli trials with parameters @xmath358 , respectively . then @xmath360 \leq \mathbb{p}\left[b(s , q ) \geq c\right ] , \ ; \text{when}\ ; c\geq sq,\ ] ] where @xmath361 and @xmath362 .
recall that a random variable @xmath363 is _ stochastically smaller _ than a random variable @xmath364 , if @xmath365\leq\mathbb{p}[v\geq t]$ ] , for all @xmath7 .
denote this by @xmath366 .
it is well known , and not so difficult to prove , ( see @xcite ) that @xmath366 if and only if @xmath367\leq \mathbb{e}[f(v)]$ ] , for every increasing function , @xmath70 , for which the expectations exist . moreover ,
the stochastic order is closed under convolutions .
the following result can be found in misra et al .
+ [ stochord ] fix @xmath368 and real numbers @xmath369 from the interval @xmath28 .
suppose that @xmath139 is a random variable from @xmath107 .
let @xmath370 be the random variable that takes values on the set @xmath371 with probabilities @xmath372 = \binom{m}{j } \mathbb{e}\left[x^j(1-x)^{m - j}\right ] , \ ; \text{for}\ ; j=0,1,\ldots , m .\ ] ] then @xmath370 is stochastically smaller than the random variable , @xmath373 , that takes values on the set @xmath371 with probabilities @xmath374 = \binom{m}{j } \mathbb{e}\left[x^m\right]^{j / m } \left(1-\mathbb{e}\left[x^m\right]^{1/m}\right)^{m - j } , \ ; \text{for}\ ; j=0,1,\ldots , m .\ ] ] see @xcite , theorem @xmath375 .
notice that the random variable @xmath373 is such that @xmath376 has the distribution of a @xmath377^{1/m}\right)$ ] random variable .
the following result is an analogue of theorem [ binbin ] that takes into account the additional information on the moments .
+ fix positive integers , @xmath112 . for @xmath31
let @xmath378 be an @xmath105-tuple of real numbers such that @xmath379 and for which the class @xmath115 is non - empty .
let @xmath29 be independent random variables such that @xmath116 , for @xmath31 .
for @xmath380 set @xmath381^{1/j}$ ] .
fix a _ positive integer _
@xmath7 such that @xmath382 . for @xmath383
let @xmath384 be the interval @xmath385 $ ] and let @xmath386 be the interval @xmath387 . if @xmath388 , for some @xmath380 , then @xmath60 \leq \min_{1\leq s\leq j}\;\left\{\frac{(st - s+1)(1-q_s)}{s(st - s+1-nq_s)}\cdot \mathbb{p}[bi(ns , q_s)\geq st - s+1]\right\ } , \ ] ] where @xmath389 , for @xmath390 .
note that @xmath391^{1/j}\}_{j=1}^{m}$ ] is an increasing sequence , for all @xmath31 .
fix @xmath392 and let @xmath357 be such that @xmath393 .
define the function @xmath394 and note that @xmath395 .
since @xmath396 , for all @xmath397 , theorem [ momopt ] implies that @xmath60 \leq \mathbb{e}\left[\max\{0 , z_{ns}-t+1\}\right ] , \ ] ] where @xmath398 and each @xmath315 takes that value @xmath399 , for @xmath400 , with probability @xmath401 = \binom{s}{\ell } \mathbb{e}\left[x_i^{\ell}(1-x_i)^{s-\ell}\right].\ ] ] from theorem [ stochord ] we know that each @xmath315 is stochastically smaller than @xmath402 , where @xmath402 is such that @xmath403^{1/s}\right)$ ] , for @xmath31 .
since @xmath49 is an increasing function , and the stochastic order is closed under convolutions , we conclude that @xmath404 \leq \mathbb{e}\left[\max\{0 , \xi_{ns}-t+1\}\right],\ ] ] where @xmath405 is the independent sum of @xmath402 s .
now @xmath406 , where @xmath407^{1/s}\right)$ ] , and so @xmath408 = \frac{1}{s}\mathbb{e}\left[\max\{0 , b_{ns}-st+s\}\right],\ ] ] where @xmath409 .
since @xmath7 is assumed to be an integer , we can write @xmath410 & = & \sum_{k = st - s+1}^{sn}(k -st+s)\cdot \mathbb{p}[b_{ns}=k ] \\ & = & \sum_{k = st - s+1}^{n } \mathbb{p}[b_{ns}\geq k].\end{aligned}\ ] ] since @xmath411 , hoeffding s theorem [ lemhof ] implies that @xmath412 \leq \mathbb{p}\left[bi(ns , q_s ) \geq k \right ] , \ ; \text{for}\ ; k\geq nsq_s , \ ] ] where @xmath413 .
summarising , we have shown @xmath60 \leq \frac{1}{s}\cdot\sum_{k = st - s+1}^{ns } \mathbb{p}[bi(ns , q_s)\geq k ] .\ ] ] finally , we use the following estimate on binomial tails ( see feller @xcite , page 151 , formula @xmath286 ) : @xmath414 \leq \frac{k - kq_s}{k - nq_s } \cdot\mathbb{p}[bi(ns , q_s)\geq k ] , \ ; \text{for } \ ; k > nsq_s , \ ] ] and the result follows . in the next section
we show that the bernstein - hoeffding method can be adapted to the case in which one has information on the conditional means of the random variables .
in this section we will work with independent , bounded random variables for which we have information on their conditional distribution .
let us be more precise after fixing some notation and definitions .
let @xmath368 be a positive integer and let @xmath415 be real numbers forming the partition @xmath416 of the interval @xmath2 $ ] ; where , for @xmath383 , we set @xmath384 to be the interval @xmath417 and @xmath418 $ ] .
now let @xmath419 be the class consisting of all random variables @xmath420 for which @xmath421= \mu_j$ ] .
formally , @xmath422= \mu_j , \ ; \text{for}\ ; j=1,\ldots , m\ } .\ ] ] finally , let @xmath423 be the class consisting of all random variables in @xmath420 for which @xmath424= q_j$ ] , i.e. , @xmath425= q_j , \ ; \text{for}\ ; j=1,\ldots , m\ } .\ ] ] suppose that we have independent , bounded random variables for which we know whether they belong to one of the above classes of random variables . in this section
we show how to employ the bernstein - hoeffding method in order to obtain bounds that take the additional information into account . in order to do so
, we will need the notion of mixture of random variables . recall that a _ mixture _ of the random variables @xmath426 is defined as a random selection of one of the @xmath427 according to a probability distribution on the index set @xmath428 .
the next result is a mixture - analogue of lemma [ coupling ] .
+ [ mix ] let @xmath289\rightarrow\mathbb{r}$ ] be a convex function .
fix positive integer @xmath368 and real numbers @xmath415 . for @xmath383 ,
let @xmath384 be the interval @xmath417 and let @xmath418 $ ] .
if @xmath139 is a random variable in @xmath1 , then there exists a random variable @xmath429 whose support is the set @xmath430 such that @xmath431=p$ ] and @xmath432\leq \mathbb{e}[f(\xi_x)]$ ]
. let @xmath433 be the event @xmath434 .
define @xmath435 to be the random variable whose distribution is the conditional distribution of @xmath139 , given @xmath433 .
it is easy to see that @xmath139 is a mixture of @xmath436 ; @xmath435 is chosen with probability @xmath437 $ ] .
now @xmath438 and so lemma [ coupling ] implies that @xmath439 , where @xmath440 is the random variable that takes that values @xmath441 and @xmath442 with probabilities @xmath443}{r_j - r_{j-1}}$ ] and @xmath444 respectively . the required random variable
can be obtained by letting @xmath429 take the value @xmath445 with probability @xmath446 $ ] , the value @xmath16 with probability @xmath447 $ ] and , for @xmath383 , the value @xmath442 with probability @xmath448+(1-\pi_{j+1})\mathbb{p}[x\in i_{j+1}]$ ] .
note that the random variable @xmath429 of the previous lemma depends on the conditional probabilities @xmath424 , j=1,\ldots , m$ ] as well as on the conditional means @xmath449,j=1\ldots , m$ ] .
so , in case we know the conditional probabilities and the conditional means of the random variables , we can find random variables that are larger , in the sense of convex order , than any random variable having the same conditional probabilities and means .
similarly , one can find a random variable that is larger , in the sense of convex order , than any random variable fromthe class @xmath419 , i.e. , when we know conditional means .
+ [ mixber ] let @xmath419 be the class defined above , corresponding to a given partition @xmath450 of the interval @xmath2 $ ] .
then there exists random variable @xmath451 that concentrates mass on the endpoints of the interval @xmath452 and @xmath386 such that @xmath453 , for all @xmath454 .
@xmath350 depends on the partition @xmath450 and the conditional means @xmath455 .
fix a random variable @xmath454 .
let @xmath435 be the random variable whose distribution is the conditional distribution of @xmath139 , given the event @xmath456 .
lemma [ mix ] implies that there exist random variables @xmath457 , that concentrate mass on the endpoints , @xmath458 , of the intervals @xmath384 such that @xmath459 .
note that , by construction , @xmath460= \mathbb{e}\left[x_j\right]= \mu_j$ ] , for all @xmath461 .
now let @xmath350 be a mixture of the random variables @xmath462 and @xmath463 ; we take @xmath350 to be equal to @xmath462 with probability @xmath464 and equal to @xmath463 with probability @xmath465 .
note that @xmath466= p$ ] .
we now show that @xmath467 .
fix a convex function @xmath289\rightarrow\mathbb{r}$ ] and let @xmath468\rightarrow \mathbb{r}$ ] be the function whose graph is the line passing through the points @xmath469)$ ] and @xmath470)$ ] .
since @xmath70 is convex , we have @xmath471 , for all @xmath472 from the interval @xmath473 $ ] . hence @xmath474 & = & \sum_{j=1}^{n}\mathbb{e}\left[f(x_j)\right]\cdot \mathbb{p}\left[x\in i_j\right]\\ & \leq & p_1 \mathbb{e}\left[f(t_1)\right ] + p_m \mathbb{e}\left[f(t_m)\right ] + \sum_{j=2}^{m-1}p_j \mathbb{e}\left[f(x_j)\right]\\ & \leq & p_1 g(\mu_1 ) + p_m g(\mu_m ) + \sum_{j=2}^{m-1}p_j \mathbb{e}\left[g(x_j)\right ] .\end{aligned}\ ] ]
since @xmath475 is linear , we have @xmath476\big ) = \mathbb{e}\left[g(x_j)\right ] , j=1,\ldots , m$ ] and so the last expression can be written as @xmath477= \mathbb{e}\left[g(x)\right ] .\ ] ] by linearity of @xmath475 , we have @xmath478 = g(p)= \mathbb{e}\left [ g(\xi)\right]$ ] . summarising , we have shown that @xmath349 \leq \mathbb{e}\left [ g(\xi)\right].\ ] ] once again , linearity of @xmath475 implies @xmath479 & = & \frac{\mu_m - p}{\mu_m -\mu_1 } \mathbb{e}\left[g(b_1)\right ] + \frac{p-\mu_1}{\mu_m -\mu_1 } \mathbb{e}\left[g(b_1)\right ] \\ & = & \frac{\mu_m - p}{\mu_m -\mu_1 } g(\mu_1 ) + \frac{p-\mu_1}{\mu_m -\mu_1 } g(\mu_m ) \\ & = & \frac{\mu_m - p}{\mu_m -\mu_1 } \mathbb{e}\left[f(t_1)\right ] ) + \frac{p-\mu_1}{\mu_m -\mu_1 } \mathbb{e}\left[f(t_m)\right ] ) \\ & = & \mathbb{e}\left[f(\xi ) \right ] \end{aligned}\ ] ] and the result follows .
the following theorem can be regarded as an improvement upon hoeffding s in the case where one has additional information on the conditional means of the random variables .
+ let the random variables @xmath29 be independent and such that @xmath480 .
fix positive integer @xmath368 and real numbers @xmath415 . for @xmath383 ,
let @xmath384 be the interval @xmath417 and let @xmath418 $ ] .
assume further that for @xmath31 there is a sequence @xmath113 such that @xmath481 = \mu_{ij}$ ] .
let @xmath482 $ ] . if @xmath7 is such that @xmath53 , then there exist @xmath483 that add up to @xmath16 , such that @xmath133 \leq \inf_{h>0}\ ; e^{-ht } \left\{\pi_1 + e^{hr_1}\pi_2 + e^{hr_{m-1}}\pi_3 + e^{hr_m}\pi_4 \right\}^n .\ ] ] the proof is similar to the proof of theorem [ maintwo ] and so we only sketch it .
lemma [ mixber ] implies that @xmath484 , for @xmath31 , where each @xmath485 concentrates mass on the endpoints , @xmath486 , of the intervals @xmath452 and @xmath386 .
hence , the arithmetic - geometric means inequality implies @xmath133 \leq e^{-ht } \left\{\frac{1}{n}\sum_{i=1}^{n } \mathbb{e}\left[e^{h\xi_i}\right ] \right\}^n.\ ] ] set @xmath487 , @xmath488 and @xmath489 , for @xmath31 .
using lemma [ mix ] , the result is obtained by setting @xmath490 , @xmath491 , @xmath492 and @xmath493 . in case one
considers random variables from the class @xmath423 , the random variable that is largest in the convex order is given by the solution of a linear program .
in particular , we have the following .
+ fix a convex and increasing function @xmath82 . fix a positive integer @xmath368 and real numbers @xmath415 . for @xmath383 ,
let @xmath384 be the interval @xmath417 and let @xmath418 $ ] .
assume further that there is a sequence @xmath494 such that the class @xmath423 is non - empty .
then there is a @xmath495 such that @xmath496 and @xmath497 \leq \mathbb{e}\left[e^{h\xi}\right ] , \ ; \text{for all } \
; x\in \mathcal{c}(p,\{i_j , q_j\ } ) , \ ] ] where @xmath25 is such that @xmath26 , i.e. , it is the optimal real such that @xmath93 = \inf_{s>0}\ ; \frac{1}{e^{st } } \mathbb{e}[e^{sb}],\ ] ] with @xmath45 .
the random variable @xmath350 depends on the solution of a linear program . since @xmath498 = p$ ] ,
the first statement is evident by lemma [ coupling ] .
let @xmath499 and set @xmath500 $ ] from lemma [ mix ] we know that , for @xmath31 , there is a random variable @xmath429 that concentrates mass on the set @xmath501 such that @xmath429 takes the value @xmath502 with probability @xmath503 , the value @xmath504 with probability @xmath505 and , for @xmath383 , the value @xmath442 with probability @xmath506 .
therefore , @xmath507 = \pi_0 e^{hr_0 } + \pi_me^{hr_m } + \sum_{j=1}^{m-1 } \pi_j e^{hr_j } , \ ] ] which implies that @xmath508 $ ] is a linear function of @xmath455 .
the required @xmath350 is obtained by maximising @xmath508 $ ] subject to the following linear constraints : @xmath509 , for @xmath380 and @xmath510 .
let us , for convenience , change a bit our notation and set @xmath127 to be the class of random variables from @xmath1 whose variance is @xmath128 . throughout this section
we will assume that @xmath128 is strictly positive .
hence @xmath511 . from proposition
[ impossible ] we know that there does _ not _ exist @xmath512 such that @xmath513 , for all @xmath514 . from lemma [ coupling ] we know that @xmath515 , for all @xmath514 but @xmath516 does _ not _ belong to the class @xmath127 , when @xmath517 . in theorem [ momopt ]
we have obtained , using bernstein polynomials , a random variable @xmath518 , that does _ not _ belong to @xmath127 , such that @xmath519 . in this section
we will construct another random variable having this property .
more precisely , we will prove the following .
+ [ xirandom ] there exists a random variable @xmath520 such that @xmath521 for all @xmath514 .
depending on the value of @xmath3 and @xmath128 , the random variable @xmath126 can yield efficiently computable bounds that are sharper than existing , well - known , bounds . after stating our main results
, we will provide some figures that illustrate the differences between the bounds . in order to construct @xmath126
we will apply lemma [ mix ] to the partition @xmath522 $ ] .
we will also need the following result that is interesting on its own .
+ [ nice ] suppose that @xmath523 are two random variables from the class @xmath127 and consider the partition @xmath522 $ ] of @xmath2 $ ] .
let @xmath429 and @xmath524 be the associated random variables given by lemma [ mix ]
. then @xmath525 \geq \mathbb{p}[\xi_y = p ] .\ ] ] assume first that @xmath526
. then @xmath527 \leq \mathbb{e}\big[f(\xi_y)\big]$ ] for the function @xmath289\rightarrow [ 0,\infty)$ ] having values @xmath528 and which is linear on the intervals @xmath529 $ ] and @xmath530 $ ] . hence @xmath531 = 1- \mathbb{p}[\xi_x = p ] \leq \mathbb{e}\big[f(\xi_y)\big ] = 1- \mathbb{p}[\xi_y = p]\ ] ] and so @xmath532 \geq \mathbb{p}[\xi_y = p]$ ] .
+ assume now that @xmath532 \geq \mathbb{p}[\xi_y = p]$ ] .
then for any convex function @xmath289\rightarrow [ 0,\infty)$ ] , we have @xmath533 and so @xmath534 -\mathbb{e}\big[f(\xi_y)\big ] & \leq & f(0)\cdot \left(\mathbb{p}[\xi_x = 0]-\mathbb{p}[\xi_y = 0]\right ) \\ & + & f(1)\cdot \left(\mathbb{p}[\xi_x = 1]-\mathbb{p}[\xi_y = 1]\right ) \\ & + & ( ( 1-p)\cdot f(0)+p\cdot f(1))\cdot \left(\mathbb{p}[\xi_x = p]-\mathbb{p}[\xi_y = p]\right ) \\ & = & f(0)\cdot \left(\mathbb{e}\big[1-\xi_x\big ] -\mathbb{e}\big[1-\xi_y\big ] \right ) + f(1)\cdot \left(\mathbb{e}\big[\xi_x\big ] -\mathbb{e}\big[\xi_y\big ] \right ) \\ & = & 0 , \end{aligned}\ ] ] where the last equality comes form the fact that @xmath535 = \mathbb{e}\big[\xi_y\big]$ ] .
consider the class @xmath127 with @xmath536 .
for every @xmath514 let @xmath537 be the random variable whose distribution is the conditional distribution of @xmath139 , given that @xmath538 . let also @xmath539 be the random variable whose distribution is the conditional distribution of @xmath139 , given @xmath540 $ ] . from lemma [ mix ]
we know that there is a random variable @xmath429 such that @xmath141=\mathbb{e}[\xi_x]$ ] and @xmath541 .
furthermore , @xmath429 is the mixture of random variables @xmath542 and @xmath543 such that @xmath544=\mathbb{e}[b_i]$ ] and @xmath545 , for @xmath546 .
in addition , @xmath542 concentrates mass on the set @xmath547 and @xmath543 concentrates mass on the set @xmath548 .
assume that @xmath429 is equal to @xmath542 with probability @xmath549 .
clearly , @xmath429 depends on @xmath139 and we now show how one can get rid of this dependence .
define @xmath126 to be the random variable @xmath550 for which @xmath551 = \min_{y\in \mathcal{b}(p,\sigma^2)}\mathbb{p}\big[\xi_y = p\big ] .\ ] ] from lemma [ nice ] we have @xmath552 , for all @xmath553 .
set @xmath554 $ ] and @xmath555-p$ ] .
off course , @xmath556 depend on @xmath139 . since @xmath557 + ( 1-\theta_x)\mathbb{e}[x_2]$ ] , we can write @xmath558 the cases in which @xmath559 and @xmath560 are both equal to zero can be excluded since they correspond to a constant random variable .
we now proceed to find the distribution of @xmath561 .
using lemma [ mix ] we compute @xmath562 & = & \theta_x \mathbb{p}[b_1 = p ] + ( 1-\theta_x)\mathbb{p}[b_2=p ] \\ & = & \frac{\ell_2}{\ell_1+\ell_2 } \left(1- \frac{\ell_1}{p}\right ) + \frac{\ell_1}{\ell_1+\ell_2 } \left(1-\frac{\ell_2}{1-p } \right ) \\ & = & 1-\frac{\ell_1\ell_2}{(1-p)p(\ell_1 + \ell_2 ) } \\ & = & 1 - \frac{1}{(1-p)p(1/\ell_1 + 1/\ell_2)}.\end{aligned}\ ] ] the last expression implies that @xmath563 $ ] is a decreasing function of @xmath559 and of @xmath560 . similarly , one can check that @xmath564 = \frac{\ell_1\ell_2}{p(\ell_1+\ell_2 ) } \quad \text{and}\quad \mathbb{p}\left[\xi_x=1\right ] = \frac{\ell_1\ell_2}{(1-p)(\ell_1+\ell_2 ) } .\ ] ] by the law of total variance we have @xmath565= \theta_x \text{var}[x_1 ] + ( 1-\theta_x)\text{var}[x_2]+ \theta_x \ell_1 ^ 2 + ( 1-\theta_x)\ell_2 ^ 2.\ ] ] hence @xmath566 or , equivalently , @xmath567 . since @xmath563 $ ] is a decreasing function of @xmath559 and of @xmath560 it follows that it attains its minimum when @xmath568 and this , in turn , implies that @xmath569 = 1-\frac{\sigma^2}{(1-p)p(\ell_1 + \ell_2 ) } .\ ] ] therefore , in order to minimise @xmath563 $ ] it is enough to solve the following optimization problem : @xmath570 elementary , though quite tedious , calculations show that the optimal solution @xmath571 equals @xmath572 therefore , the required random variable @xmath126 has the following distribution : * if @xmath573 , then @xmath126 takes the values @xmath574 and @xmath16 with probability @xmath575 , @xmath576 and @xmath577 respectively . *
if @xmath578 , then @xmath126 takes the values @xmath574 and @xmath16 with probability @xmath579 , @xmath580 and @xmath581 , respectively . * if @xmath582 , then @xmath126 takes the values @xmath574 and @xmath16 with probability @xmath583 , @xmath584 and @xmath585 , respectively .
the proof of theorem [ xitheorem ] is an application of lemma [ xirandom ] .
it is very similar to the proof of theorem [ maintwo ] and theorem [ yuyipr ] and so we briefly sketch it .
the first statement can be proven in the same way as theorem [ maintwo ] .
the second statement follows from the fact that @xmath586 and by looking at the smallest positive integer , @xmath267 , that is @xmath587 .
as in theorem [ yuyipr ] , we can find an @xmath588 such that the function , @xmath235 , that is equal to @xmath445 for @xmath589 and , for @xmath590 , it is a straight line passing through the points @xmath226 and @xmath591 satisfies @xmath208 and @xmath592\leq \mathbb{e}\left[f\left(\sum_i \xi_{p_i,\sigma_i}\right)\right],\ ] ] for a supposedly optimal function @xmath49 with @xmath218 .
it is not easy to find a closed form of the bound given by theorem [ xitheorem ] .
nevertheless , the bound can be easilly implemented .
note that the previous bound concerns functions from the class @xmath81 .
we end this section by performing some pictorial comparisons between several bounds discussed in this article . before doing so ,
let us bring to the reader s attention the following , well - known , bound that is due to bennett @xcite .
bennett s approach was simplified by cohen et al .
@xcite . in particular , by employing the bernstein - hoeffding method to the exponential function , cohen et al .
have shown the following . + fix positive integer @xmath5 and assume that we are given a pair @xmath593 for which the class @xmath127 is non - empty .
let @xmath29 be independent random variables such that @xmath594 , for @xmath31 .
. then @xmath20 \leq \left\{\left(\frac{\alpha}{\beta}\right)^{\beta } \left ( \frac{1-\alpha}{1-\beta}\right)^{1-\beta } \right\}^n,\ ] ] where @xmath596 and @xmath597 .
see @xcite .
+ + + our numerical experiments suggest that , when @xmath128 is not very small , the bound given by theorem [ xitheorem ] is tighter than bennett s bound .
note that we can also apply the bound given by theorem [ momopt ] to random variables from the class @xmath127 ; it is not difficult to implement this bound . in order to build a concrete mental image
let us fix the parameter @xmath3 and consider random variables @xmath598 such that @xmath599 in a similar way as in proposition [ yuyiprtwo ] one can show that it suffices to consider the infimum , in the bound of theorem [ xitheorem ] , over the set @xmath600 .
we can now put the computer to work to calculate the bound @xmath601.\ ] ] figure [ fig : compare ] shows comparisons between bennett s bound , the bound obtained in theorem [ momopt ] and the bound of theorem [ xitheorem ] .
the abscissae in these figures correspond to the variance .
notice that , when the variance is large , the bounds given by theorems [ momopt ] and [ xitheorem ] are sharper than the bennett bound . in the next section we stretch a limitation of the bernstein - hoeffding method .
so far we have employed the bernstein - hoeffding method to sums of independent and _ bounded _ random variables .
the reader may wonder whether the method can be employed in order to obtain bounds on deviations from the expectation for sums of independent , non - negative and _ unbounded _ random variables .
we will show , in this section , that in this case the method yields a bound that is the same as the bound given by markov s inequality .
let us remark that this fact was already known to hoeffding ( see the footnote in @xcite , page @xmath182 ) but we were not able to find a proof ; we include a proof for the sake of completeness .
hence the case of non - negative and unbounded random variables requires different methods and the reader is invited to take a look at the work of samuels @xcite , @xcite , @xcite and feige @xcite for further details and references .
the case of non - negative and unbounded random variables seems to be less investigated than the case of bounded random variables .
talagrand ( see @xcite , page 692 , comment @xmath602 ) already mentions that it is unclear how to improve hoeffding s inequality without the assumption that the random variables are bounded from above .
let us fix some notation .
+ for given @xmath603 , let @xmath604 the class of non - negative random variables whose mean equals @xmath605 .
formally , @xmath606=\mu\ } .\ ] ] now , for @xmath31 , fix @xmath607 and @xmath608 .
if @xmath609 , then one can estimate @xmath610 \leq \frac{1}{f(t ) } \mathbb{e}\left [ f\left(\sum_{i=1}^{n}x_i\right)\right],\ ] ] where @xmath49 is a non - negative , convex and increasing function .
a crucial step in the bernstein - hoeffding method is to minimise the right hand side of the last inequality with respect to @xmath49 .
we may assume that we minimise over those functions @xmath70 for which @xmath218 .
we now show that this minimisation leads to a bound that is the same as markov s .
note that markov s inequality yields @xmath611\leq \frac{\sum_i \mu_i}{t}$ ] .
recall the definition of the class @xmath81 , from the introduction , and let @xmath612 be the class consisting of all functions @xmath87 such that @xmath218 . in this section
we report the following .
+ for @xmath31 , let @xmath427 be the random variable that takes the values @xmath445 and @xmath7 with probabilities @xmath614 and @xmath615 , respectively . clearly , we have @xmath616 .\ ] ] in a similar way as in theorem [ yuyipr ]
one can show that @xmath617 = \inf_{\varepsilon \in [ 0,t ) } \mathbb{e}\left[\max\left\{0 , \frac{\sum_i y_i -\varepsilon}{t-\varepsilon}\right\}\right].\ ] ] since @xmath618 , a similar argument as in proposition [ yuyiprtwo ] shows that the optimal @xmath247 in the right hand side of the last equation is equal to @xmath445 .
therefore , @xmath619=
\frac{1}{t } \mathbb{e}\left [ \sum_{i=1}^{n } y_i \right ] = \frac{1}{t } \sum_{i=1}^{n } \mu_i\ ] ] and the result follows .
hence , in the case of non - negative and unbounded random variables , the method can not yield a bound that is better than markov s bound . +
* acknowledgements * the authors are supported by erc starting grant 240186 `` migrant , mining graphs and networks : a theory - based approach '' .
we are grateful to xiequan fan for several valuable suggestions and comments .
a. cohen , y. rabinovich , a. schuster , h. shachnai , ( 1999 ) .
_ optimal bounds on tail probabilities : a study of an approach _ , advances in randomized parallel computing , comb .
optim . , vol .
5 , kluwer acad . publ . , p. 124 .
s. samuels , ( 1968 ) _ more on a chebyshev - type inequality for sums of independent random variables _ , purdue stat .
mimeo , ser .
s. samuels , ( 1969 ) .
_ the markov inequality for sums of independent random variables _ , annals of math .
40 ( 6 ) , p. 19801984 . | we show that the bernstein - hoeffding method can be employed to a larger class of generalized moments .
this class includes the exponential moments whose properties play a key role in the proof of a well - known inequality of wassily hoeffding , for sums of independent and bounded random variables whose mean is assumed to be known . as a result we can generalise and improve upon this inequality .
we show that hoeffding s inequality is optimal in a broader sense .
our approach allows to obtain `` missing '' factors in hoeffding s inequality whose existence is motivated by the central limit theorem .
the later result is a rather weaker version of a theorem that is due to michel talagrand . using ideas from the theory of bernstein polynomials ,
we show that the bernstein - hoeffding method can be adapted to the case in which one has information on higher moments of the random variables .
moreover , we consider the performance of the method under additional information on the conditional distribution of the random variables and , finally , we show that the method reduces to markov s inequality when employed to non - negative and unbounded random variables .
_ keywords _ : hoeffding s inequality , convex orders , bernstein polynomials |
at the ilc , some beam parameters have to be measured at a permille level precision to fully exploit the physics potential of machine and detectors @xcite . however
, contrary to luminosity and beam energy measurements for which similar precisions were already achieved at previous colliders , this accuracy is unprecedented for measurements of the beam polarisation .
the polarisation measurement at the ilc will combine the measurements of two dedicated polarimeters , located upstream and downstream of the @xmath0 interaction point , and data from the @xmath0 annihilations themselves .
while @xmath0 annihilation data will finally provide an absolute scale , the polarimeters allow for fast measurements , can give feedback to the machine , reduce systematic uncertainties and add redundancy to the system @xcite .
both upstream and downstream polarimeters will use compton scattering of high power lasers with the electron and positron beams @xcite .
circularly polarised laser light hits the particle bunches under a small angle and typically in the order of 1000 electrons are scattered per bunch .
the energy spectrum of the scattered particles depends on the product of laser and beam polarisations , so that the measured rate asymmetry w.r.t .
the ( known ) laser helicity is directly proportional to the beam polarisation .
figure [ fig : comptonproc](a ) shows the compton cross section versus scattered electron energy for a beam energy of @xmath1 gev and a photon energy of @xmath2 ev .
the large polarisation asymmetry is clearly visible near 25.2 gev , the compton edge energy .
in addition , this edge energy hardly depends on the beam energy as can be seen in figure [ fig : comptonproc](b ) .
( 14.0 , 5.5 ) ( 0.00 , 0.00 ) ( 7.10 , 0.00 ) ( 0.00 , 0.00 ) ( a ) ( 7.20 , 0.00 ) ( b ) since the electrons scattering angle in the laboratory frame is below @xmath3rad , a magnetic spectrometer is used to transform the energy spectrum into a spacial distribution and lead the electrons to the polarimeter s cherenkov detector .
it will consist of staggered ` u - shaped ' aluminum tubes lining the tapered exit window of the beam pipe as illustrated in figure [ fig : ildpol - chertube](a ) .
the tubes will be filled with a high - threshold cherenkov gas ( @xmath4 ) and are read out by photodetectors ( pds ) .
compton - scattered electrons traversing the tubes u - bases generate cherenkov radiation which is reflected upwards to the photodetectors [ fig : ildpol - chertube](b ) . developing a cherenkov detector suitable for achieving the aforementioned precision demands improvements in various areas . in order to study the entire experimental setup ,
a two - channel prototype cherenkov detector has been designed , simulated , constructed , and operated successfully in laboratory and beam tests .
( 14.0 , 9.1 ) ( 0.00,-0.20 ) ( -0.10 , 0.00 ) ( 0.00 , 8.50 ) ( a ) ( 0.00 , 3.50 ) ( b ) the same right - handed coordinate system ( see figure [ fig : ildpol - chertube ] ) is used throughout this note : the beam travels in positive @xmath5 direction , the @xmath6-axis points upwards , and the @xmath7-axis to the left when looking in beam direction .
various requirements driving the design of the actual ilc cherenkov detectors have also been taken into account for the prototype detector and its construction : * efficient & homogenous light response to the primary @xmath8 flux ( high reflectivity also + for short wavelengths ; geometry / surfaces to illuminate pd cathode homogeneously ) * gas- and light - tightness ( control linearity ; stabilize response over macroscopic times ) * thin inter - channel walls ( avoid loosing electrons ; avoid background creation ) * robustness w.r.t . backgrounds ( high - threshold gas avoids low - energetic electron + background ; good layout keeps pds / calibration source outside the beam plane ) * calibration system ( monitor response in - situ ; indep . of beam availability )
the last two items lead to the idea of u - shaped channels . with increasing length of the u - basis more cherekov light
is produced , but the alignment requirements become more stringent and additional reflections will decrease the light yield again .
simulations show that a length of 15 cm yields a sufficient amount of light while introducing only one additional reflection under a glancing angle .
contrary to the ilc - like design of staggered channels , the prototype detector consists of only two parallel , non - staggered channels ( figure [ fig : simgeant ] ) , but it still allows to test all relevant aspects of the full detector .
for the design of the prototype detector and the interpretation of the testbeam data , an optical simulation based on geant4 @xcite has been created .
the prototype is simulated according to the technical drawings and is surrounded by a wide box filled with perfluorobutane ( @xmath4 ) as cherenkov gas .
figure [ fig : simgeant ] shows the internal channel structure with the electron beam passing from left to right through the u - basis of the right - hand side ( 8.0 , 5.0 ) ( 0.50,-0.40 ) detector channel . although the surrounding base box is not shown , the ambient cherenkov gas still encompasses the entire structure .
this leads to cherenkov radiation emitted outside the aluminium tubes , i.e. before the beam enters and after it exits the internal channel structure .
however , light emitted outside the channel volumes can not reach the photodetectors mounted on the hind u - leg .
care was taken to properly simulate the inner channel walls by introducing two different types of aluminium with average refractive indices of @xmath9 for the diamond - milled aluminium making up 3 of the 4 inner walls , and @xmath10 for the two 0.15 mm - thin rolled foils comprising the middle wall between both channels .
the wavelength dependence of both refractive indices is implemented by interpolating linearly between different wavelengths for which the reflectivities had previously been measured .
the purpose of the optical simulation is to determine some key figures such as the photon yield / electron , the average number of reflections and possible asymmetry effects due to the chosen geometry or utilized materials .
this simulation ends at the top of the hind u - leg , right before the produced photons would hit the pd cathode .
figure [ fig : cherspec - lightyieldsim](a ) shows the wavelength spectrum of these photons with its typical @xmath11 dependence directly after the optical simulation ( dashed grey line ) and convoluted with a typical quantum efficiency ( solid black line ) .
the applied quantum efficiency is shown in the inlet in figure [ fig : cherspec - lightyieldsim](a ) and belongs to the 2@xmath122 multi - anode photomultiplier ( r7600u-03-m4 ) from hamamatsu @xcite .
this photodetector is one of two multi - anode photomultipliers ( mapms ) chosen at the very beginning of the design phase based on earlier stand - alone studies of different photodetector types @xcite . due to the layout and square geometry of these multi - anode photomultipliers ,
the channel cross section was chosen to be [email protected] mm@xmath13 .
( 14.0 , 6.5 ) ( -0.05 , 0.00 ) ( 3.10 , 2.90 ) ( 6.90 , 0.00 ) ( 0.00 , 0.00 ) ( a ) ( 7.20 , 0.00 ) ( b ) to study the influence of the detector geometry and the reduced reflection capability of the inner foil wall on the expected light yield on the pd cathode , two scenarios are simulated with the same beam parameters ( 2d gaussian profile , @xmath14 1.5 mm ) , but different refractive indices for the inner foil wall . while the light yield proved to be symmetric about the central @xmath7 and @xmath5-axes
when all four channel surfaces are simulated with the same reflectivity , a clear @xmath7-asymmetry when the inner foil wall is simulated with the correct reduced reflectivity , see figure [ fig : cherspec - lightyieldsim](b ) .
overall , the light yield is higher at larger @xmath7 and lower at smaller @xmath7 allowing to conclude that on average the photons are reflected an uneven number of times under a glancing angle off the channel walls . a simple calculation using a cherenkov angle of approximately @xmath15 and an average light path length of 175 mm ( half the cherenkov length plus the length of the u - leg ) gives an estimate of one single reflection off the channel walls . the two narrow bands at @xmath16 mm , where the light yield is also visibly reduced originate from the @xmath17 reflection at the end of the cherenkov section of a channel .
they also appeared when all channel walls were simulated with the same reflectivity .
the observed light yield asymmetry has also been quantified in the simulation with a grid - scan consisting of 4@xmath124 equidistant points ( left - hand side detector channel and 10,000 electrons per shot ) . due to the geometry of the channel layout
, the @xmath6-position of the beam entry point on the channel u - basis translates directly to the @xmath5-position on the anode readout plane of the photodetector .
however , the results of the simulated scans in the @xmath7 and @xmath6 ( @xmath5 ) directions are not displayed here , but are included in figure [ fig : asymxz - data](a , b ) on page as a comparison with the respective asymmetries calculated from measured beam test data .
after the final assembly of the prototype detector , its basic functionality was checked in various short - term laboratory tests where the gas- and light - tightness was also confirmed . in spring 2009 , the prototype was taken to the elsa electron accelerator in bonn and set up in one of the external beam lines directly behind a dipole magnet bending the electron beam towards a dump set into the floor .
the detector was mounted on its turnable base plate and additionally affixed to a stage moveable along the @xmath7- and @xmath6-axis .
two angles could only be adjusted approximately : @xmath18 defining a tilt about the @xmath7-axis ( in the beam - plane ) and @xmath19 defining a tilt about the @xmath5-axis in a plane orthogonal to the beam - axis .
while the latter was easily adjusted using a water - level , the former tilt angle @xmath18 was more difficult to adjust , since the detector had to be positioned at an angle of about @xmath20 with respect to the horizontal to match the downwards slope of the electron beam line .
elsa comprises three stages : injector linacs , a booster synchrotron and the stretcher ring with a circumference of 164.4 m which leads to a turn time of 548 ns for relativistic electrons@xcite . although the overall beam structure due to the rf acceleration is fixed ( 274 phase space buckets distributed evenly across the ring )
, the actual number of electron bunches per turn can be ajusted via partial filling .
in addition , the beam spot can be focused to about 1 - 2 mm and the extraction current is tuneable from approximately 10 pa to slightly above 200 pa .
both these features lead to numerous electrons traversing the u - basis of the detector channels simultaneously , emitting cherenkov photons and producing large and well pronounced cherenkov signals . during the test period for the prototype detector
, elsa was operated in _
booster mode _ with the electrons being injected at an energy of 1.2 gev and subsequently accelerated to 2.0 gev . since
beam extraction is not possible during the refill / acc .
phase , the beam is extracted in intervalls of @xmath21s for every @xmath22s cycle ( ratio @xmath234:1 ) . since no specific triggers were employed , the beam clock signal looped through a function generator was used to provide the necessary gate for the readout electronics ( a charge sensitive analog - to - digital converter , or qdc ) .
the gate length was adjusted such that the detector integrated over all electron bunches of one complete elsa turn .
one of the first measurements performed with beam is shown in figure [ fig : m4-diffbc - tiltscan](a ) .
the extraction current was varied from lower to higher values and the corresponding cherenkov signals reproduce this behaviour .
another important observation is the stability of the dark current ( dc ) rate .
the hatched histogram in figure [ fig : m4-diffbc - tiltscan](a ) represents the dark current ( dc ) rate recorded without beam , while the open histograms show cherenkov signals for different extraction currents . for each open histogram ,
the area underneath the rightmost peak is about four times larger than the area underneath the corresponding dark current peak originating from the refill / acc .
phases without beam extraction ( ratio @xmath234:1 ) .
the characteristic dc rate depends primarily on the applied bias voltage and on temperature , but not ( at least not directly ) on the beam current .
however , changes in beam conditions usually also influence other parameters .
thus , a stable dc rate shows that no effects from temperature or beam background are discernible so far .
( 14.0 , 6.2 ) ( -0.10 , 0.00 ) ( 7.10 , 0.00 ) ( 0.00 , 0.00 ) ( a ) ( 7.30 , 0.00 ) ( b ) the detector position and horizontal tilt are aligned using direct beam data .
typically , for a tilted detector the maximal cherenkov signal at a given beam position is smaller than for a perfectly aligned detector . while one scan in vertical direction is sufficient to adjust the @xmath6 position , two scans ( one on each channel ) are preferable . to determine the correct tilt in the ( @xmath24)-plane
, the channel front face is scanned horizontally by the electron beam for six different tilt angles @xmath25 .
for all scans , i.e. different tilt angles , the highest observed signals are compared as shown in figure [ fig : m4-diffbc - tiltscan](b ) .
the best alignment is achieved for a tilt angle of @xmath26 ( represented by the blue square ) .
this proceedure improves the accuracy on the horizontal tilt angle from @xmath27 to less than @xmath28 .
it is also possible to extract some of the detector dimensions , such as the channel width or the distance between channel centers , from beam data .
a comparison of these derived values and the given prototype specifications can be used to disentangle different effects that are either connected to the prototype detector itself , to its alignment w.r.t . the beam axis , or maybe exclusively to the beam .
figure [ fig : m4xscanbelong - r7400xscansbround](a ) shows @xmath7-scan data recorded with the 2@xmath122 mapm ( r7600u-03-m4 ) for an ellipsoidally elongated beam spot .
two gaussian fits indicate the respective channel centres to be at @xmath29 mm and @xmath30 mm , leading to a distance of @xmath31 mm .
the nominal distance of @xmath32 mm is given by the sum of the channel width and the thickness of the inner foil wall .
both values agree rather well considering that effects from different beam spots , the reduced reflectivity of the inter - channel wall , or residual misalignment were not yet taken into account .
( 14.0 , 6.6 ) ( -0.10 , 0.00 ) ( 7.10 , 0.00 ) ( 5.00 , 1.15 ) ( 10.70 , 1.15 ) ( 0.00 , 0.00 ) ( a ) ( 7.30 , 0.00 ) ( b ) figure [ fig : m4xscanbelong - r7400xscansbround](b ) shows the @xmath7-scan data recorded with the r7400u-06 single - anode photomultiplier ( sapm ) for a nearly round beam spot .
from two sigmoidal fits to the edges of the signal and at a 75%-level of the maximal signal strength the effective channel width is measured to be @xmath33 mm .
again , the agreement with the nominal width of @xmath34 mm is very good , indicating that the prototype is functional and behaves as expected .
the fact that the 2@xmath122 mapm data in figure [ fig : m4xscanbelong - r7400xscansbround](a ) does not exhibit such a clear plateau as the sapm data is assmued to be due to the different beam profiles during the respective measurements .
+ another measurement series uses the 8@xmath128 mapm ( r7600 - 00-m64 ) , whose anode is more finely segmented and allows studying the intra - channel distribution of the light yield .
unfortunately , two out of eight qdc readout channels were broken during the beam test period , so that two quadrants of the 8@xmath128 mapm ( i.e. 32 anode pads ) needed to be grouped into the six remaining readout channels .
figure [ fig : m64-ac6-xyscan ] shows ( a ) the chosen anode readout configuration for the 8@xmath128 mapm , ( b ) data from an @xmath7-scan across both detector channels versus the beam @xmath7-position , and ( c ) the same data , but taking into account the different anode groupings for both channels and measuring the distance of the beam entry point w.r.t .
the opposite channel wall . ( 14.0 , 5.0 ) ( -0.10 , 0.30 ) ( 2.90 , 0.00 ) ( 8.60 , 0.00 ) ( 0.00 , 0.00 ) ( a ) ( 3.00 , 0.00 ) ( b ) ( 8.70 , 0.00 ) ( c ) as can already be seen from figure [ fig : m64-ac6-xyscan](b ) the highest light yield is always measured on the anode pad opposite the respective beam entry point .
for example , if the beam enters the channel u - basis at the @xmath7-position of anode 2 , the highest light yield is measured on anode 3 and vice versa .
this can be seen even more clearly in figure [ fig : m64-ac6-xyscan](c ) , where all anode responses exhibit the same triangular shape .
the data clearly confirm the simulation result of ( on average ) one additional reflection under a glancing angle off the channel walls apart from the unavoidable @xmath35 reflection from u - basis to u - leg .
any remaining amplitude differences still visible in figure [ fig : m64-ac6-xyscan](c ) is either due to different anode sensitivities , residual detector misalignment , or a mixture of both effects .
+ the asymmetries in figure [ fig : asymxz - data](a , b ) are derived from @xmath7 and @xmath6-scan data for the left - hand side detector channel and are directly comparable to the simulation . for each beam position the corresponding asymmetries @xmath36 and
@xmath37 are calculated by subtracting the light intensity measured in the left ( lower ) channel half from the intensity measured in the right ( upper ) half : @xmath38 where @xmath39 ( @xmath40 ) correspond to the intensities in the left ( lower ) half of a channel and @xmath41 ( @xmath42 ) to the intensities in the right ( upper ) half , respectively .
the non - vanishing value of the simulated @xmath7-asymmetry at @xmath43 in figure [ fig : asymxz - data](a ) originates from the different reflectivities of the inner channel walls ( diamond - milled aluminium versus foil reflectivity ) .
( 14.0 , 6.5 ) ( -0.10 , 0.00 ) ( 7.10 , 0.00 ) ( 0.00 , 0.00 ) ( a ) ( 7.30 , 0.00 ) ( b ) two @xmath7-asymmetries are calculated from the beam data : one for anodes 4 + 5 ( one for anodes 7 + 6 ) located at lower ( higher ) @xmath5-values atop the lower ( upper ) half of the left detector channel .
the two @xmath5-asymmetries are defined for the orthogonal grouping of anodes 5 + 6 ( and anodes 4 + 7 ) located atop the left ( right ) part of the channel at higher ( lower ) @xmath7-values , respectively .
( see figure [ fig : m64-ac6-xyscan](a ) for details on the readout configuration . ) both asymmetries , in @xmath7 and in @xmath6 direction , are clearly observable in beam data demonstrating that the effect is not washed out by imperfections of the channel surfaces not modelled in the simulation .
this means that also the asymmetries could be used for alignment and calibration purposes in the final ilc cherenkov detectors if position sensitive photodetectors are employed .
qualitatively the measured asymmetries agree well with those from simulation .
especially the @xmath7-asymmetries are not point - symmetric , but exhibit clear offsets in @xmath7 , as expected due to the lower reflectivity of the thin inter - channel wall .
quantitatively however , some of the measured asymmetries deviate from the ideal expectation .
the @xmath7-asymmetry from anodes 7 + 6 exhibits a different slope , while the @xmath5-asymmetry from anodes 4 + 7 seems to be shifted .
these differences hint towards residual misalignment around the @xmath6 and/or @xmath5-axes and the influence of the non - perfect beam profile also still needs to be taken into account .
a two - channel prototype cherenkov detector has been designed , simulated and successfully operated under beam conditions .
the optical simulation of the prototype detector is based on geant4 and provides the light distribution at the photodetector cathode .
the modular design of the prototype allows fast exchanges of the photodetector and calibration modules and thus facilitates operation during beam tests .
so far a number of expectations from simulation could already be confirmed with a first preliminary analysis of some of the beam data .
one example of a successful application of simulation methods to real data , is the extraction of intra - channel position information from various scans .
presentation : + ` http://ilcagenda.linearcollider.org/contributiondisplay.py?contribid=56&sessionid=6&confid=4175 ` ilc global design effort and world wide study , _
`` international linear collider reference design report - volume 2 and 3 '' , _ august 2007 .
moortgat - pick _
et al . _ ,
* 460 * ( 2008 ) 131 , [ arxiv : hep - ph/0507011 ] .
s. boogert , a.f .
hartin , m. hildreth , d. kfer , j. list , t. maruyama , k. mnig , k. moffeit , g. moortgat - pick , s. riemann , h.j .
schreiber , p. schler , e. torrence , m. woods , jinst * 4 * ( 2009 ) p10015 .
the geant4 collaboration , nucl .
instrum . meth .
. res . a * 506 * , issue 3 ( 2003 ) , 250 - 303 ; the geant4 collaboration , ieee trans . nucl .
science * 53 * no . 1 ( 2006 ) , 270 - 278 ; issn : 0018 - 9499 c. bartels ,
c. helebrant , d. kfer , j. list , _ poster : `` photodetector studies and prototype simulation of a cherenkov detector for an ilc polarimeter , '' _ ieee 2008 nuclear science symposium and medical imaging conference , dresden , germany , october , 2008 .
c. helebrant , `` evaluation of different photodetector types for an ilc polarimeter , '' + nucl .
instrum .
res . a * 610 * , issue 1 ( 2009 ) , 387 - 389 . c. helebrant , ph.d .
`` in search of new physics using polarisation hera and ilc , '' _ + university of hamburg , hamburg , germany , 2009 . | precise knowledge of all beam parameters is crucial to fully exploit the physics potential of the international linear collider ( ilc ) . a sufficiently accurate measurement of the beam polarisation
can only be achieved using dedicated high energy compton polarimeters combined with well - designed arrays of cherenkov detectors .
this note focuses on the design and detailed simulation of a suitable cherenkov detector prototype and provides an overview of first results from a highly successful beam test period . |
recently a significant progress has been achieved in the nanofabrication and design of new technologies for very high density storage on hard disks ( densities above 500 gbit / in@xmath0 ) and magnetoresistive random access memory ( mram ) . for the further miniaturization of the data storage and the magnetic memory
it is important to decrease the size of nanoelements .
magnetic defects in such nanoelements storing bits of information can have very small size .
the properties of skyrmion in thin metallic magnetic films ( the small size about 1 nm and stability ) may allow to use their two - dimensional ( 2d ) arrays in magnetic storage devices . to the best of our knowledge
a magnetic memory based on skyrmions was not developed yet .
there are no analogs of storage of so extremely high density .
skyrmions are strongly localized scale invariant analogs of magnetic bubble domains ( mbd ) @xcite .
the axially symmetric skyrmion solution for boson fields minimizing the hamiltonian and describing the structure of nuclear matter was first found by skyrme @xcite .
belavin and polyakov based on nonlinear @xmath1-model have established that skyrmion is a topologically nontrivial metastable minimum of energy @xcite . since then
the properties of skyrmions in two - dimensional magnets and thin magnetic films have been widely investigated in last few decades .
slow relaxation spin dynamics of skyrmion was studied by abanov and pokrovsky .
they calculated the radius of the skyrmion as a function of time .
they have found that the interaction of the ruderman - kittel - kasuya - yosida ( rkky ) type together with the uniaxial magnetic anisotropy can stabilize the skyrmion . in quasi-2d antiferromagnets the skyrmions
were indirectly observed by f. waldner by the esr @xcite and the heat capacity measurements @xcite in a good agreement with the belavin and polyakov ( bp ) theory .
the appearance of skyrmions in the qhe was predicted by sondhi et al . .
in the regime of fractional qhe skyrmions were discovered experimentally in gaas quantum wells by means of optically pumped nmr and by measurements of the knight shift , which is proportional to the electron magnetization . for integer
qhe skyrmions were detected in thermal magnetotransport measurements @xcite and in optical absorption @xcite .
recently another topological excitations , magnetic vortices were observed and studied experimentally in permalloy and co nanodisks by means of the lorentz microscopy and magnetic force microscopy ( mfm ) .
the evolution and stability of such magnetic vortices in a small isotropic cylindrical magnetic dot was analyzed theoretically in .
the equilibrium size of skyrmion in 2d thin magnetic films is extremely small .
this is one of serious obstacles for its experimental observation . for soft magnets the size of vortices
is defined by competition between the exchange and magnetic dipole interaction . in uniaxial magnets ,
strong uniaxial anisotropy together with the fourth order exchange interaction leads to small size of skyrmion .
recent great progress in nanotechnology gives a chance to detect experimentally skyrmions in thin magnetic films .
the experimental methods employed for studying skyrmions in qhe systems and vortices in soft magnetic dots can be also applied to 2d uniaxial magnets .
so far the only technique potentially capable of atomic resolution is the spin - polarized scanning tunneling microscopy ( spstm ) . due to stability of the skyrmion in an uniaxial thin metallic magnetic film it is possible to create it by a strong magnetic tip . in this paper
we study the properties of the mbd created by the tip as well as its relaxational dynamics when tip moves away uniformly and vertically from the film .
when the distance between the film and the tip exceeds a critical value , it is not able to confine the mbd anymore and the latter collapses and gradually transforms into a stable skyrmion . for numerical simulations we use a stabilization mechanism provided by the interplay between next nearest neighbor antiferromagnetic interaction and the magnetic anisotropy .
we also show that the demagnetization energy of the skyrmion is equivalent to the renormalization of the single ion anisotropy constant .
we find that demagnetizing field produced by a single skyrmion is strongly localized around skyrmion center .
the plan of this paper is as follows . in section [ model ]
we introduce a theoretical model and equations of motion .
we calculate the demagnetization energy of the skyrmion in section [ demag ] .
section [ tip ] is devoted to the magnetic field generated by a cylindrical tip . in section
[ static ] we study static properties of the mbd and the skyrmion solving numerically landau - lifshitz equation with the fitting method . in section
[ discrete ] we study analytically and numerically a discrete spin model with the dipolar interaction and find stability condition for the static skyrmion .
relaxational dynamics of the transformation of a mbd into a skyrmion in the presence of an uniformly moving tip is considered in section [ relax ] .
our conclusions and discussions are presented in the last section [ conclusion ] .
we consider the two - dimensional heisenberg model with the exchange interaction up to the fourth order derivatives and uniaxial anisotropy in continuous approximation .
its hamiltonian has a following form @xmath2 where @xmath3 and @xmath4 here the energy density @xmath5 consists of the exchange energy @xmath6 , the uniaxial anisotropy energy @xmath7 ; @xmath8 contains the interaction energy of cylindrical tip with the magnetic film and the demagnetization energy . in our notations
@xmath9 is the exchange interaction constant , @xmath10 is the easy - axis anisotropy constant , @xmath11 is a positive dimensionless constant , characterizing the fourth order exchange interaction ( we consider just this term because of its simplicity ) , @xmath12 is the spontaneous magnetization of the film , @xmath13 is the magnetic field created by the magnetic tip ( cylinder ) at the surface of the film as a function of the polar radius @xmath14 and the distance @xmath15 between the cylinder and the plane of the film ; @xmath16 is the film thickness . in a thin magnetic film
the demagnetization energy of a localized magnetic excitation with the radius @xmath17 at @xmath18 simply renormalizes the anisotropy constant @xmath10 ( @xmath19 ) . in the opposite case @xmath20
the demagnetization energy is more complicated .
below we calculate it for the skyrmion . for iron the exchange length @xmath21 . therefore , an ultrathin iron film consists of few monolayers .
the dynamics of a classical 2d ferromagnet is governed by the landau - lifshitz equation ( lle ) for the magnetization vector @xmath22 @xmath23}{\delta { \bf m}({\bf r},t ) } + \nonumber \\ { \gamma \ ,
\dot{\bf m}({\bf r},t ) \times { \bf m}({\bf r},t ) \over m_0 } \label{lle}\end{aligned}\ ] ] here @xmath24 ; @xmath25 and @xmath26 are the gyromagnetic ratio and the bohr magneton , respectively , and @xmath27 is the dimensionless gilbert damping constant . the lle equation ( [ lle ] ) in angular variables @xmath28 and @xmath29 can be derived from the lagrange - hamilton variational principle with the following lagrangian density @xmath30 @xmath31 in the static case the lle eq .
( [ lle ] ) read : @xmath32 the right - hand side of eq .
( [ stlle ] ) originates from a constraint : @xmath33 at any point @xmath34 .
the classical ground state is double degenerate due to uniaxial magnetic symmetry .
a topologically nontrivial minimum of the hamiltonian ( [ h ] ) corresponds to the skyrmion solution @xmath35 the skyrmion is characterized by its radius @xmath36 and an angle @xmath37 of the global spin rotation ; @xmath14 and @xmath38 are polar coordinates
. for isotropic magnets the energy of skyrmion does not depend neither on @xmath36 nor on @xmath37 .
in real systems the degeneracy with respect to @xmath36 is lifted by uniaxial magnetic anisotropy and by the fourth order exchange interaction .
another static solution of the lle is the well known mbd , which was the subject of intensive study during last few decades .
here we consider an mbd with the topological charge one .
the structure of such a soliton can be approximately described by the `` domain - wall '' ansatz : @xmath39 where @xmath40 is the mbd radius .
since we consider an ultrathin magnetic film , the mbd exists only in the presence of an external magnetic field and collapses without this field .
) versus polar radius normalized to the film thickness @xmath41 .
solid and dashed lines correspond to @xmath42 and @xmath43 , respectively . the radius of skyrmion @xmath44 . ] let us now return to the demagnetization energy .
the magnetostatic problem for the non - topological mbd was solved first by thiele @xcite .
he expressed @xmath45 in terms of elliptic integrals .
the thiele s solution for demagnetizing field @xmath46 can be approximated with surprisingly good accuracy by @xmath47 where dimensionless radius @xmath48 and dimensionless energy @xmath49 are introduced .
for magnetostatic interaction the topological charge does not play any role .
hence the thiele s solution persists when dipolar forces are taken in account . for ultrathin magnetic films consisting of few layers
the ratio of the exchange length to the thickness is @xmath50 .
therefore , the mbd solution is unstable without external magnetic field .
the reading of the information stored by skyrmions can be performed by the measurement of the skyrmion magnetic field .
below we calculate this field and its energy .
the fourier transform of the skyrmion magnetic field has the following form @xmath51 the magnetostatic energy in terms of the magnetization fourier - components reads : @xmath52 to calculate the magnetic field we substitute the fourier transform of the skyrmion magnetization ( [ skyrm ] ) into ( [ demagfield ] ) and perform the inverse fourier transformation .
the result is as follows : @xmath53 \nonumber \\
h^{d}_{r } = 4 \pi m_0 \ , r_{s}^2 \int_{0}^{\infty } f_{r}(q , z , h ) \ , k_{0}(q r_s ) \ , j_{1}(q r ) \ , q dq , \label{skyrmdemag}\end{aligned}\ ] ] where @xmath54 the origin of the frame of reference coincides with the skyrmion center .
the skyrmion demagnetizing field for @xmath55 is shown in fig .
[ demag ] . the demagnetization energy of the skyrmion ( [ demagskyrm])in units @xmath56 has the following form : @xmath57 where the last term is a part of magnetostatic energy renormalizing anisotropy constant : @xmath58 and @xmath59 has the form @xmath60 where @xmath61 , @xmath62 , @xmath63 and @xmath64 are the complete elliptic integrals of first and second kind , respectively , and @xmath65 is the generalized hypergeometric function .
we consider ultrathin film @xmath66 .
employing an appropriate expansions for the elliptic integrals and hypergeometric functions we find @xmath67 the expression ( [ expanddemag ] ) can be replaced by @xmath68 in the interval @xmath69 with the accuracy of half percent .
the spin - polarized scanning tunneling microscopy ( sp - stm ) and spectroscopy ( sp - sts ) offers a great potential to unravel the structure of a single skyrmion and skyrmion arrays on the nanometer scale .
) and ( [ hzhr ] ) at a distance @xmath15 from the film : @xmath70 0 , 2 , 5 and 10 .
@xmath71 is the anisotropy field of the film and @xmath72 is the magnetic length .
solid and dashed lines correspond to @xmath73 and @xmath74 components , respectively .
the radius of magnetic tip @xmath75 , the ratio of magnetization of the tip to that of the film is @xmath76 . ] consider a semi - infinite cylindrical magnetic tip .
the stray magnetic field of the cylindrical tip is well known .
its @xmath77 and @xmath14 components are @xmath78 where @xmath79 , @xmath80 and @xmath15 are the magnetization , the radius of cylinder and the distance between the tip and magnetic film plane respectively .
the radial component of the magnetic field of the tip @xmath81 at the edge @xmath82 can be expressed in terms of the hypergeometric function : @xmath83 this value diverges logarithmically when the distance between the tip and the magnetic film approaches zero : @xmath84 in a special case when the distance between the film and the tip @xmath85 the components of the magnetic field are : @xmath86 , \\
h_r(r ) & = & \pi m_t \ , \left({r \over r}\right)^{\sigma}\ , _ { 2 } f_{1 } \left({1 \over 2 } , { 3 \over 2 } , 2 , \left({r
\over r}\right)^{2 \sigma}\right ) , \label{hzrd0}\end{aligned}\ ] ] where @xmath87 and @xmath88 is the heaviside function .
the expression ( [ hzrd0 ] ) logarithmically diverges at the distances close to the tip s edge @xmath82 @xmath89 the components of magnetic field of the tip versus @xmath14 at different @xmath15 are depicted in fig .
[ htip ] . at large distances @xmath90
the magnetic tip is equivalent to a magnetic charge @xmath91 and asymptotically the magnetic field is coulomb - like : @xmath92 to make tip stronger by increasing magnetic charge of the tip @xmath93 it is possible to increase the radius of the tip or its magnetization .
the first method is more effective .
at @xmath85 and different values of the ratio of the tip magnetization to that of the film @xmath94 . a ) profile of @xmath95 for @xmath96 .
the radial component of magnetic field is shown on the same figure .
b ) a skyrmion - like solution strongly distorted by the cusp of @xmath74 at the edge of the tip at @xmath97 .
the energies of mbd @xmath98 and skyrmion @xmath99 versus @xmath94 are shown in insert .
@xmath100 is the energy of the unperturbed skyrmion . ]
[ d0 ] in order to study the relaxation dynamics of the skyrmion created by the magnetic tip it is necessary first to analyze static solutions of the lle for the magnetic film .
a rough estimate of the minimal tip radius stabilizing the mbd at very small distance @xmath15 can be extracted from the balance of energy : @xmath101 , where @xmath102 is the linear tension of the domain wall . for more accurate analysis
the knowledge of the magnetization and magnetic field distribution is necessary .
several cylindrically symmetric soliton - like solutions of the lle have been found .
they are the static skyrmion solution , the dynamic skyrmion with the spin precession and the slowly precessing magnetic bubble domain ( mbd ) .
the precessing solitons can live longer than solitons without precession , though they finally also collapse due to dissipation , in accordance with the derick - hobart theorem @xcite .
the collapse does not proceed if the skyrmion configuration is stabilized by the higher order exchange interaction .
stabilization also occurs in ferromagnets without center of inversion .
let us rewrite the lle in terms of spherical coordinates @xmath28 and @xmath29 , which determine the direction of magnetization : @xmath103 where @xmath104 , @xmath38 is the azymuthal angle in the film plane ; @xmath105 is the magnetic length and @xmath106 is the ferromagnetic resonance frequency . in the static solution
@xmath107 and @xmath108 obeys an ordinary differential equation : @xmath109 where @xmath110 .
equation ( [ theta ] ) should be solved with the following boundary conditions : @xmath111 , @xmath112 .
one of the most powerful tools to solve the nonlinear ordinary differential equations is the fitting method .
the heart of this method is the matching of a numerically derived solution to a known asymptotics at small and large distances from the center of nonlinear excitation by varying the shooting parameter .
the key problem in this method is to find a proper trial function , with known asymptotics at @xmath113 and @xmath114 and fits it to the asymptotics of the real solution .
we solve lle for @xmath108 ( [ theta ] ) by the fitting method .
we choose the trial function @xmath108 in the following form : @xmath115 .
\label{trial}\ ] ] the trial function ( [ trial ] ) has two advantages .
first , it interpolates between two principal magnetic textures : the mbd and the skyrmion by varying the fitting parameter @xmath116 , and second , it has a valid asymptotic at @xmath113 .
being distorted by the radial component of the tip magnetic field @xmath74 , the radial component of the magnetization acquires an unusual power asymptotics @xmath117 .
this behavior is more pronounced when the tip is close to the magnetic film ( @xmath118 ) . in this case
the radial component of the tip magnetic field can not be treated as a perturbation , since it causes a strong distortion of the skyrmion .
small and large values of @xmath119 correspond to the skyrmion solution ( [ skyrm ] ) and the mbd , respectively .
we match the numerical solutions of lle with the known asymptotic at infinity .
let us now consider a special case when the distance between the surface of the and the tip @xmath15 is zero .
the angle @xmath120 at large distances decreases as @xmath121 , similarly to @xmath122 . at small distances ,
the solution for @xmath28 has the skyrmion asymptotic .
we have solved numerically equation ( [ theta ] ) for @xmath28 with the shooting method for different @xmath123 . for each value of @xmath124
we have found two different values of the fitting parameter @xmath119 satisfying the matching condition .
we treat the solution with larger @xmath119 as the mbd , whereas the solution with smaller @xmath119 is interpreted as the skyrmion - like solution .
profiles of @xmath108 for different values of @xmath124 are presented in fig .
the cusp of @xmath122 component of the magnetic field of the tip produces the noticeable distortion of skyrmion solution in the region of the tip s edge .
this distortion grows with @xmath124 increasing as it is seen from fig .
strongly distorted skyrmion - like excitation is the second solution of landau - lifshitz equation , but its energy is larger than that of the mbd .
thus , the magnetic tip nucleates mbd at static conditions .
[ d0 ] illustrates that the radius of the generated mbd equals roughly to the radius of the tip .
this result can be obtained banalytically by the direct minimization of energy .
( [ freeen ] ) as a function of mbd radius @xmath40 at different distances between the tip and film .
curves with @xmath125 , @xmath126 and @xmath127 correspond to @xmath128 = 0.1 , 0.2 , 0.5 respectively .
the ratio of the radius of the tip to the magnetic lenth @xmath129 and @xmath130 . ] in order to calculate the tip - film interaction energy we substitute the expressions ( [ hzhr ] ) for @xmath73 and @xmath74 and equation ( [ mbd ] ) into ( [ h ] ) . using approximate relations : @xmath131 and @xmath132
, we find @xmath133 where @xmath134 , @xmath56 is the unperturbed skyrmion energy : @xmath135 ; and @xmath136 .
the total energy of the mbd @xmath137 in units @xmath56 is represented by the sum of 4 terms @xmath138 where @xmath139 , @xmath140 , @xmath141 and @xmath142 are the exchange , the anisotropy , zeeman and demagnetization energies , respectively .
the partial sum @xmath143 can be represented by an interpolation formula : @xmath144 valid with good precision in a broad interval of the ratio @xmath145 . the first term in equation ( [ exchanis ] )
is the linear tension of the domain wall , the second is the correction to it due nonzero topological charge of the mbd .
the zeeman energy is given by eq .
( [ wz ] ) .
the demagnetizing energy @xmath146 for an ultrathin magnetic film ( @xmath147 ) slightly renormalizes the exchange energy and can be neglected in the first approximation .
this approximation is correct provided the radius of the mbd is much larger than the thickness of the film .
the energy of mbd versus its radius @xmath40 for different @xmath15 is shown in fig .
[ bubenergy ] .
if @xmath148 , the energy minimum at @xmath149 disappears ; the tip can not confine mbd anymore and the latter shrinks .
accepting @xmath150 , we find from eq .
( [ freeen ] ) the range of parameters , in which the nucleation of the mbd is favorable : @xmath151 where @xmath152 , @xmath153 @xmath154 \label{f}\ ] ] @xmath155 , @xmath156 ; @xmath157 and e(k ) are the complete elliptic integrals of the first and second kind , respectively .
the total energy of the mbd has the following form @xmath158 where @xmath159 and @xmath37 is the global rotation angle . in the equlibrium state @xmath160 . at a fixed radius of the tip @xmath80 , the bubble nucleates when the distance between the tip and the film becomes less than some critical value @xmath161 .
is the distance between the tip and the surface of the film ; @xmath80 is the tip radius , @xmath130 , @xmath162 . ]
in this section we study the dynamics of the bubble creation and its transformation into a stable skyrmion solving numerically time - dependent landau - lifshitz equation on a discrete lattice .
we employ the model of a classical two - dimensional heisenberg ferromagnet with the nearest , next nearest ( nnn ) , and next to next nearest neighbor ( nnnn ) interaction and dipolar interaction .
the corresponding hamiltonian reads : @xmath163 where the summations run over all spin sites @xmath164 , its nearest neighbors @xmath165 , nnn along diagonals @xmath166 and nnnn @xmath167 .
we assume the ferromagnetic exchange @xmath168 between nn , antiferromagnetic exchange @xmath169 between nnnn ; the sign of nnn exchange @xmath170 is not fixed ; @xmath171 denotes easy axis anisotropy constant and @xmath172 is the dipolar long range coupling constant . both exchange interactions @xmath170 and @xmath169 mimic the real rkky interaction in metallic ferromagnets mediated by conduction electrons . in usual magnets
the dimensionless ratio @xmath173 is of the order @xmath174 .
fourier transform of the oscillatory rkky interaction has the form @xmath175 , where @xmath176 is the density of states and @xmath177 is the lindhardt function , whose expansion into the tailor series over powers of @xmath178 is @xmath179 since both coefficients at @xmath180 and @xmath181 , have the same sign , the ferromagnetic state provided by this interaction satisfies the condition of stability ( see section [ model ] ) . in the continuous limit
the hamiltonian ( [ dish ] ) takes a following form @xmath182 where @xmath183 , @xmath184 is the demagnetization field energy .
the perturbative part of the hamiltonian @xmath185 is @xmath186 where @xmath187 is the scaled magnetic field .
the skyrmion is stable in the continuous model if both @xmath188 and @xmath189 are positive .
the discreteness changes the topology and make the continuous decay of skyrmion possible . to study the discreteness effect in continuous model haldane has proposed to remove one plaquette in the center of the skyrmion @xcite .
an explicit calculation of the skyrmion decay by this mechanism was performed in the work .
the skyrmion is metastable in a discrete lattice if in a configuration with only one reversed spin its rotation to the parallel orientation requires overcoming a potential barrier .
this requirement is satisfied in a square lattice if @xmath190 . though important for simulation on a discrete lattice
, this effect does not play any role for real itinerant magnets , in which spin density is continuous .
as we have shown earlier , the skyrmion stability requires @xmath189 to be positive , whereas the sign of @xmath191 remains indefinite .
to study the dynamic transformation of the mbd into the skyrmion we solved the time - dependent lle by the runge - kutta method for a sample with free boundaries .
the sample had a circular shape and contained 11308 spins located at the sites of a square lattice .
we choose time step 10@xmath192 for good convergence . for dynamical simulations we generated initially the mbd imitating a strong tip in a close vicinity of the magnetic film .
the radius of the generated mbd is approximately equal to the radius of the tip ( the difference between radii of the tip and the mbd is below 5 % ) .
once the bubble was generated , we started to move away the tip uniformly ( @xmath193 ) .
the mbd gradually transformsed into a stable skyrmion or collapsed depending on the exchange parameters @xmath170 and @xmath169 .
the range of the exchange parameters corresponding to the skyrmion stability is represented in fig .
[ jjstab ] .
the skyrmion stability criterion is in good agreement with the numerical data . and
@xmath194 : crosses correspond to a stable skyrmion , curve 1 corresponds to @xmath195 , curve 2 to @xmath196 .
simulations are in a good agreement with the criterion of stability against long - wave spin perturbations . ] the spin structure of the mbd is shown in fig .
[ bubble ] .
the shape of the bubble is slightly distorted during the collapse .
this distortion owes to the effect of the square lattice . .
] we also studied the difference between continuous and discrete approximation by numerical simulations with different anisotropy constants @xmath171 .
the discrete models approaches the continuous approximation when anisotropy constant @xmath171 decreases and the domain wall width @xmath197 increases ( see table [ table1 ] ) .
.normalized parameters of the soliton versus anisotropy constant @xmath171 derived from numerical simulations .
all parameters are normalized to ferromagnetic exchange interaction @xmath168 .
antiferromagnetic exchange interaction constant @xmath198 and @xmath199 .
[ cols="<,^,^,^,^,^,^,^ " , ] [ table1 ] the process of transformation passes three stages . in the first stage the bubble domain is controlled by the removing tip until it reaches the distance @xmath200 . during the second stage
the mbd shrinks under the action of the linear tension .
when its radius reaches the value @xmath201 , it can not be more described as the mbd and acquires the profile of a skyrmion .
a rough interpolation formula for the transformation time @xmath202 following from semi - intuitive consideration reads : @xmath203 in this equation the first term is the time during which the tip reaches the distance @xmath204 from the film . at larger distance
the interaction between the tip and the film is negligibly small .
the second term corresponds to the shrinking of the mbd under the action of the laplace force from the domain wall .
the laplace force is equal to @xmath205 , where @xmath206 is the linear tension .
the local velocity of the domain wall is antiparallel to the radius and equal to @xmath207 , where @xmath208 is the magnetic viscosity .
the time @xmath209 necessary to decrease significantly the radius is equal to integral @xmath210 this gives exactly the second term in equation ( [ totaltime ] ) . finally , the third term is the relaxation time for the skyrmion derived in .
this simple interpolation fits the simulation data with very good accuracy .
a more sophisticated description of the transformation process can be obtained by substitution of the interpolation between the mbd and skyrmion given by equation ( [ trial ] ) into the hamiltonian ( [ h])and integrate it over all variables . the resulting expression is the hamiltonian for slow variables @xmath80 and @xmath37 . the dimensionless lagrangian density @xmath211 can be constructed given the hamiltonian density @xmath137 by the following ansatz : @xmath212 where @xmath213 , @xmath214 and @xmath17 is the radius of topological excitation . to incorporate the dissipation it is possible to use the rayleigh dissipative function @xmath215 . in dimensionless variables it reads @xmath216 we will not present the equations of motion here to avoid lengthy formulas with no essentially new results
. they can be useful when a more detailed description of the process will be in order .
for real magnetic film with @xmath217 , tip with radius @xmath218 , damping constant @xmath219 and tip s velocity @xmath220 transformation time is @xmath221 .
the equilibrium size of the skyrmion is @xmath222 where @xmath223 $ ] and @xmath224 is the lattice spacing .
skyrmion spin structure obtained by numerical simulations of time - dependent landau - lifshitz equation is represented in fig .
[ skyrmion ] . an effective magnetic field corresponding to the exchange interaction
@xmath225 is @xmath226 .
for @xmath227 it is about 89 t and for @xmath228 it is 112 t. for existing strong magnetic tips real value of created magnetostatic field is about 1 t. therefore we consider the ratio @xmath229 . for iron film and tip with radius 50 nm minimum magnetic field of the tip @xmath230 .
, the exchange interaction constants are @xmath231 , @xmath232 , @xmath233 . ]
the main result of this work is the numerical evidence that the bubble domain of the 50 nm size initiated by a strong magnetic tip in a thin magnetic film shrinks into a very small , few nanometers metastable skyrmion . for this purpose
we employed the mc and the runge - kutta methods on the square lattice .
the static skyrmion and the magnetic bubble domain were found numerically as solutions of the static landau - lifshitz equation .
the role of the magnetic tip is two - fold .
its magnetic field stabilizes the bubble domain , which otherwise is unstable in a thin film .
being removed it does not confine the mbd anymore and the latter starts to shrink conserving its topological charge .
the skyrmion is the final result of this shrinking .
thus , the position of the skyrmion is determined with high accuracy by the tip .
this is important for the experimental observation of the skyrmion , since it is very small and its magnetic field is strongly localized as it was established by our calculations . on the other hand ,
strong localization of the skyrmion magnetic field makes possible to read the information stored by skyrmions if the distances between them and between them and empty spaces are more than @xmath234 , where @xmath36 is skyrmion radius .
such a distance is sufficient to ensure that skyrmions do not perturb the magnetization around .
limitations on exchange constants accepted in our work stem from the requirement of the skyrmion stability .
one of them is associated with the discrete character of the model and is not essential for a real itinerant magnet with continuous distribution of magnetization .
another one occurs in continuous model : the part of energy proportional to the fourth power of the gradient must be positive .
being imposed onto a lattice model it leads to an inequality , which provides sufficiently large antiferromagnetic exchange between the nnnn . in real itinerant
ferromagnets the natural rkky interaction ensures the stability .
thus , the skyrmion stability makes it is possible to create it by circular magnetic tip and to observe it by spin - polarized scanning tunnelling microscopy having a nanometer resolution .
it is not yet clear how the creation of the next skyrmion by lateral scanning tip will influence an already existing skyrmion .
when this problem will be solved , the skyrmion arrays could be a promising candidate for the magnetic memory of unprecedented high density above 10 tbit / in@xmath235 .
this work has been supported by nsf under the grants dmr0072115 and dmr0103455 , by doe under the grant fg03 - 96er45598 and by the telecommunications and information task force at texas a&m university . | skyrmions in thin metallic ferromagnetic films are stable due to competition between the rkky interaction and uniaxial magnetic anisotropy .
we mimic the rkky interaction by the next - nearest - neighbors ferromagnetic and antiferromagnetic exchange interactions .
we demonstrate analytically and numerically dissipative transformation of a bubble created by a strong cylindrical magnetic tip into a stable skyrmion . |
understanding when and how the universe re - ionized is fundamental to our understanding of how galaxies and large scale structure form and evolve and is sensitive to global cosmological parameters . in particular ,
the _ fraction of neutral hydrogen _
, @xmath8 , in the intergalactic medium ( igm ) is closely tied to early galaxy formation because it is related to the gas accretion rate onto galaxies . from current measurements
it is still unclear when re - ionization occurred and what the sources of re - ionizing radiation are .
the best of such current measurements come from cosmic microwave background ( cmb ) experiments and high - redshift quasar studies , with additional constraints from lyman break ( lbg ) and ly@xmath0 emitting ( lae ) galaxy studies .
wmap @xcite and planck @xcite place a @xmath9 constraint on when re - ionization occurred , based on the optical depth to the cmb due to thompson scattering of electrons .
these data are usually fit by a quick re - ionization at @xmath10 , but are also fully consistent with a more gradual re - ionization with a tail ending at @xmath11 @xcite .
direct measurements of the optical depth from quasars indicate that the universe is neutral up to @xmath12 , based on the highest redshift quasars known today @xcite .
furthermore , ultraviolet ( uv ) continuum measurements of lbgs between @xmath13 @xcite suggest that galaxies have a difficult time re - ionizing the universe until later times unless the luminosity function is unusually steep at the faint end , or the continuum escape fraction is high @xcite .
the fraction of strong ly@xmath0 emitters within lbg samples should give us a more direct , complementary , and unique measurement of @xmath8 and therefore how quickly and when the universe is re - ionizing .
fundamentally , ly@xmath0 photons are scattered in areas where the igm contains more neutral hydrogen , so the escape fraction of ly@xmath0 photons is proportional to the volume of re - ionized hydrogen around the young galaxies .
hence the fraction of galaxies with strong ly@xmath0 emission is related to the neutral fraction of the igm @xcite .
however , it is important to note that this probe is also sensitive to the evolution of the interstellar medium ( ism ) inside galaxies ( like dust , see @xcite ) , so one must understand the effects of galaxy evolution to probe the igm .
the ly@xmath0 emission of lbg galaxies ( selected using broad bands ) is indicative of re - ionization ending at @xmath11 and a neutral hydrogen fraction of @xmath14% at @xmath15 @xcite .
in particular the fraction of strong ly@xmath0 emitters in lbg samples is found to rapidly drop beyond @xmath16 over a range of @xmath17 , a timescale of only @xmath18 myrs @xcite .
an alternative to lbg selection is the use of narrow - band ( nb ) filters to directly detect laes at specific redshifts ( e.g. , @xcite and references therein ) .
this method allows one to directly map the ly@xmath0 lfs as a function of redshift , which can then be compared to the lbg uv continuum lfs to estimate the neutral igm fraction .
an overall change in the ly@xmath0 lfs between @xmath19 has been firmly established by large samples of spectroscopically confirmed laes @xcite .
but the source of this change could be either an evolution in the igm or a change in the internal ism of the galaxies .
the evolution of the ly@xmath0 lf based on laes beyond @xmath20 is far less clear .
apart from a few spectroscopically confirmed laes at @xmath15 ( one spectroscopically confirmed out of two at @xmath21 @xcite and one spectroscopically confirmed out of three at @xmath22 @xcite ) , there are no confirmed laes at higher redshifts . a total sample of @xmath23 candidate laes at @xmath24 is known @xcite .
@xcite and @xcite favor a non - evolution of the ly@xmath0 lf between @xmath3 ( see also @xcite ) , which is in tension with other narrow band searches for laes at @xmath25 that only place limits on the number counts of laes @xcite .
the reason for this tensions may be low - redshift interlopers and false detections in the lae samples . at @xmath26 ,
both of these are estimated to contribute less than 1020% ( see e.g. , @xcite ) , at higher redshifts , these contribution are not known , yet , but are probably much higher ( see @xcite and this work ) .
spectroscopic follow - up observations of high redshift candidate laes are therefore necessary to resolve the tensions between the lae and lbg results at @xmath20 and to constrain the process of re - ionization at higher redshifts . in this work ,
@xmath27 , and h@xmath28 km s@xmath29 mpc@xmath29 . ] , we present keck - i mosfire spectroscopic follow - up of two @xmath30 lae candidates ( @xmath31 ) originally found by @xcite .
we then go on to compare these results to existing data at lower redshift ( @xmath32 ) and to an empirical model derived from the lbg uv continuum lf and observed equivalent width distribution ( @xmath33 ) .
this allows us to place limits on the neutral fraction of the igm at @xmath34 ( @xmath35 ) and enables us to predict the lae lf at @xmath36 ( @xmath37 ) .
the two targets of our study are among the brightest lae candidates at @xmath30 ( 12.1 and 8.6 @xmath38 10@xmath39 erg / s/@xmath40 , respectively , measured in unb filters assuming negligible continuum ) .
these targets were initially selected and published by @xcite and throughout this work we refer to these as lae1 ( brightest ) and lae2 ( second brightest ) , respectively . both laes were detected with an ultra narrow band ( unb ) filter in the cosmos field @xcite using newfirm @xcite .
details of the data reduction and selection are given in @xcite , but we give a brief summary of their results here .
the effective surface area of the unb survey is @xmath41 760 arcmin@xmath42 .
the unb filter used for these candidates is centered at a wavelength of 1.056@xmath43 m and has a width of @xmath44 .
this is a dark region of the spectra between bright night sky lines , and selects objects with ly@xmath0 emission at a redshift @xmath30 .
the unb data were acquired over a course of a year in three different sets of observations ( february 2008 , february & march 2009 ) .
this means transient objects with periods of @xmath45 year were rejected ( see @xcite and later in this section ) .
the total usable observations add up to @xmath41100 hours distributed over 32 nights , resulting in a limiting magnitude ( defined as the 50% completeness limit ) of 22.4 ab in the unb filter .
the area used to select these objects is covered by a second unb filter centered at 1.063@xmath43 m with the same width as well as deep ground based broad - band data from subaru in the optical ( g , b , v , r , i , z ) and ukirt and vista in the nir ( y , j , h , k ) .
this allows one to exclude continuum on the blue and red side of the potential ly@xmath0 emission line and should have eliminated low - z interlopers .
both of the candidates are not detected in any of the broad band filters as well as the second unb filter .
this results in rest - frame equivalent width lower limits of @xmath46 and @xmath47 for lae1 and lae2 , respectively .
we observed the two lae candidates ( @xmath0 = 10@xmath4800@[email protected] , @xmath51 = + [email protected] @xmath0 = 10@xmath4800@[email protected] , @xmath51 = + [email protected] ) with the mosfire @xcite spectrograph on the keck - i telescope on the nights of january 15 & 16 , 2013 .
each candidate was observed with a separate mask created using the mosfire automatic gui - based mask application ( magma , version 1.1 ) and aligned using bright 2mass stars .
the conditions were photometric on both nights , with an average seeing around @xmath53 .
the observations were carried out in @xmath54-band ( @xmath55 ) using the yj grating and a @xmath56 slit width resulting in a resolution @xmath57 .
we used 180s exposures with 16 multiple correlated double samples .
the telescope was nodded by @xmath58 with respect to the mask center position between exposures .
the total integration times were @xmath59 for lae1 and @xmath60 for lae2 , respectively . before creating the mask
, we verified that the 2mass , cosmos , and newfirm astrometric systems agreed to within measurable errors ( @xmath61 ) . during the observations we make sure that the masks were properly aligned by using either alignment stars and/or bright filler targets .
in addition several bright sources with known fluxes and morphologies from the zcosmos - bright spectroscopic survey @xcite were placed on the mask to verify slit losses ( estimated to be 40 - 50% ) .
we observed 12 and 4 of these galaxies in the lae1 and lae2 masks , respectively .
the comparison of the expected spatial position from magma to the final spatial position on the reduced 2d spectra indicates that the alignment was better than @xmath62 during the observations .
we used the public mosfire python data reduction pipeline for sky subtraction , wavelength calibration , and co - addition of the single exposures .
the pipeline performs an a
b / b a subtraction and co - adds the single exposures using a sigma - clipped noise weighted mean after shifting them to a common pixel frame and masking bad pixels . the atmospheric oh sky lines are used for wavelength calibration .
the final 2d spectra have a spatial resolution of 0.18@xmath63 per pixel and a spectral resolution of 1.09 @xmath64 per pixel .
figure [ fig : slit ] shows the final 2d spectra ( degraded to @xmath65 ) together with the slit positions on sky .
we measured an rms noise of @xmath66erg / s/@xmath40 ( 4.4@xmath64 resolution element ) in the 10545 - 10565@xmath64 wavelength region , in good agreement with the estimated noise from the the mosfire exposure time calculator , corrected for our estimated slit losses .
absolute flux was measured using the white dwarf spectrophotometric standard star gd71 .
the standard star was observed during the same nights with identical settings and reduced in the same way as the science exposures .
we present the sensitivity curve for the mosfire @xmath54-band in figure [ fig : sensitivity ] together with the line fluxes of the two targets derived from unb filters .
this shows that we would have clearly detected the two laes as it is further discussed below . assuming the observed fluxes given in @xcite at 10560@xmath64 and based on our measured noise and our seeing of 1@xmath63
, we expect to detect the two sources at a signal - to - noise of 12.4 and 8.2 , respectively , with a line width of 200km / s ( e.g. , @xcite ) .
even with a seeing as bad as 2@xmath63 , the expected signal - to - noise is still 8.8 and 5.8 , respectively . to verify the snr calculation and lack of detection , we simulated the expected 2d spectra by adding lines to the reduced 2d spectra .
for these simulations we assumed the total measured flux was distributed over a truncated gaussian with a rest - frame fwhm ranging from @xmath67 ( observed from stacked spectra it is @xmath68 , e.g. , @xcite ) . for the spatial extent
we assume a gaussian with fwhm of 1@xmath63 corresponding to our seeing .
the following results of our simulation are not sensitive to the actual spatial extent .
we find that the total flux of such a line would have to be less than @xmath69 erg / s/@xmath40 for not to be visible in our data ( for the range in rest - frame line fwhm ) .
vice versa , to miss lae1 ( lae2 ) in our data , we would require a rest - frame fwhm of more than 10@xmath64 ( 7@xmath64 ) .
figure [ fig : slit ] shows the simulated spectra rescaled to @xmath70 assuming the line fluxes measured by @xcite , a line rest - frame fwhm of 1.5@xmath64 , and spatial extent of 1@xmath63 .
this shows a clear expected detection of both ly-@xmath0 lines .
our firm non - detection of line emission in the targeted laes yields an upper limit in ly@xmath0 line flux of @xmath71 erg / s/@xmath40 .
we therefore rule these candidates out on a 7 and 5@xmath72 level , respectively .
this puts on solid ground a recent less significant non - detection by @xcite in 7.5 hours of lbt observation with the luci nir spectrograph .
given these limits , the sources must either be a transient event with decay times of @xmath73 year , very short periodic ( @xmath74 year ) with a large change in flux , or artifacts and/or noise spikes in the data .
considering transients , the most likely events with similar rates are super - luminous supernovae ( ssne ) or agns .
low redshift sne are favorable because the rest - frame nir emission is decaying less rapidly than the optical @xcite .
these events can account for the magnitude change measured in the unb filters @xcite .
however , a simple calculation suggests that a @xmath75 ssn is visible for maximum @xmath76 days ( observed ) including the rise of luminosity before its peak .
however , @xcite searched for objects with variability on these time scales and removed them , therefore we believe these are an unlikely source of contamination , although up to three such events could have happened within 0.2 deg@xmath42 during the year of observations depending on imf @xcite . furthermore , short period agns can be excluded as a source of contamination , because of the amplitude of the variability which exceeds that in known agns @xcite
we thus conclude that the detections are most likely artifacts and noise .
there are several reasons why this could happen .
first of all , detections near the edges of an image can be caused by enhanced noise .
also , estimates of the limiting magnitude by using 50% completeness simulations and/or the use of inappropriate aperture sizes with respect to the seeing may lead to false detections . in the case of the @xcite candidates , the authors use 50% completeness simulations to estimate their limiting magnitudes . also , their candidates seem to lie systematically close ( @xmath77 arcmin ) to the chip gaps between the four newfirm arrays . combined with the findings of @xcite and @xcite , who also find no real detections ,
this raises significant questions about the reliability of the narrow band filter technique with nir detectors for detecting laes at @xmath20 .
note that for @xmath26 , where large spectroscopic follow - up studies of lae candidates are possible , the fraction of low - z interlopers and spurious objects is usually @xmath78% .
whatever the reason , the non - detection of lae1 and lae2 in the mosfire spectra places important limits on the lae lf and implies strong evolution of it at @xmath79 as it will be discussed in the following section .
-band 1@xmath72 sensitivity per 4.4@xmath64 resolution element is shown .
the measured sensitivity is consistent with that of the exposure time calculator corrected for slit losses .
we should be able to detect the two lae candidates at several @xmath72 as shown by the red symbols representing their line fluxes as measured in the unb filter by @xcite.[fig : sensitivity ] ]
a large number of studies have looked at the ly-@xmath0 luminosity function at @xmath26 .
a summary of the surveys at @xmath81 is given in table [ tab : lf1 ] and the mean data points adopted for this redshift range are shown in figure [ fig : literature ] , panel a. it can be seen that the lae lf changes only slightly in this redshift range .
schechter functions fitted to the data as a function of redshift result in less than 15% change in @xmath82 and @xmath83 , respectively @xcite . note that in this and the following comparisons of lfs , we account for eventual differences in the cosmologies assumed by the authors .
furthermore , some authors apply a correction to their ly@xmath0 luminosities to account for absorption of the ly@xmath0 forest .
this correction is debated as it is shown recently that the ly@xmath0 line profile is asymmetric at @xmath84 where igm absorption is negligible .
this suggests that ly@xmath0 is already redshifted when escaping the galaxy and most probably make the above correction factor superfluous and result in an overestimation of the lae luminosity @xcite .
the lfs presented in this paper are not corrected by this factor . at @xmath85
there are severals major studies @xcite .
all of these use the subaru / suprime - cam camera with the nb812/nb921 filters .
additional constraints come from @xcite compiling a large sample of lae surveys .
the above studies are summarized in table [ tab : lf2 ] .
these large studies have significant disagreements in the derived luminosity functions with the various studies citing contamination rates , selection functions , and spectroscopic incompleteness as possible sources of disagreement .
the @xcite study uses several widely spaced fields to rule out cosmic variance as the source of the discrepancy . in figure
[ fig : literature ] we combine the various studies , and find that while the fits to the lae lfs done by the different authors disagree , the data are consistent within errors , indicating counting statistics and fitting methods are the likely source of the discrepancy .
we adopt weighted averages of the data points for the redshifts @xmath86 , @xmath87 , and @xmath88 as indicated by the colored symbols in figure [ fig : literature ] panels a through c. in panel d we also show lae detections by @xcite , @xcite , @xcite fluxes , broad - band magnitudes , and ews ) and we therefore include them here . ] , and @xcite at @xmath15 with their weighted averages .
we note that there are differences in the normalization of the above studies which are likely linked to sample incompleteness and contamination ( estimated to be less than 20% for these studies ) .
these uncertainties are captured in the individual error bars , which we take into account in the final error bars of the weighted averages . at @xmath30 , we combine the candidate detections from @xcite and @xcite with the two remaining candidates from @xcite by adding up the comoving volumes of the studies . the new limits at @xmath30 are shown in figure [ fig : literature ] panel e. the single points are shown in figure [ fig : literature ] panel d together with the limit from @xcite shown as gray line .
finally in figure [ fig : literature ] panel e , we show our combined luminosity functions over the redshift range @xmath89 .
this clearly shows a rapid evolution in the number density of bright laes at @xmath90 .
however , it is unclear whether this evolution is driven by changes in the igm opacity , or evolution in the density of the underlying galaxy population .
we will disentangle these two effects in the following section .
ly@xmath0 emission is produced in young galaxies with a substantial amount of on - going star - formation .
it is therefore the amount of uv radiation and the ism of a galaxy which constrains the amount of ly@xmath0 emission .
as the ly@xmath0 photons escape from the galaxy , they get scattered in areas of dense neutral hydrogen in the igm .
the amount of neutral hydrogen around galaxies sets the amount of ly@xmath0 emission that can be measured by our telescopes .
as soon as galaxies are formed , they start to re - ionize larger and larger bubbles of neutral hydrogen around themselves and the transparency for ly@xmath0 photons is increased . by recording the amount of ly@xmath0 emission , i.e. , the rest - frame equivalent width ( ew@xmath91 ) distribution , as a function of redshift , it is therefore possible to estimate the change in the volume fraction of neutral hydrogen , @xmath8 , and therefore map the re - ionization process .
however , the change in the fraction of ly@xmath0 emitting galaxies also depends on the density of the underlying galaxy population as well as on internal ( ism ) properties of the galaxies , like star formation rate and dust content .
studies of the ly@xmath0 emission properties of _ uv - continuum selected lbgs _ suggest that the ly@xmath0 emission is rising with redshift in galaxies at @xmath92 @xcite , where the universe is thought to be fully re - ionized .
in particular , @xcite note that the ew distribution in this redshift range ( @xmath93 ) is skewed to larger rest - frame ew values for higher redshifts .
this suggests evolution of the internal properties of galaxies ( e.g. , dust , @xcite ) enhancing the amount of ly@xmath0 emission with increasing redshift ( e.g. , @xcite ) . in order to constrain the fraction of neutral hydrogen at @xmath34
, we have to separate these effects from the igm .
we therefore first model the intrinsic ( i.e. , without igm absorption ) ly@xmath0 lf ( @xmath944.1 ) .
later , we will compare this intrinsic lf to the observed lfs at different redshifts ( @xmath95 ) and , combined with two possible implementations of the re - ionization process , constrain @xmath8 ( @xmath35 ) . to separate ism from igm effects on the ly@xmath0 lf
( see also @xcite ) , we first create an empirical model of the lae lf based on the uv lf and the ly@xmath0 rest - frame equivalent width ( ew@xmath91 ) distribution at @xmath96 , where the igm is fully re - ionized . in brief , we assume the @xmath97 uv - continuum lfs of lbgs derived by @xcite and @xcite . these lfs can be well explained by assuming that the luminosity and stellar mass of a galaxy is directly related to its dark - matter halo assembly and gas infall rate @xcite .
especially the lf at @xmath20 are therefore put on more solid ground .
we then convolve these uv lfs with two observed ly@xmath0 ew@xmath91 distributions of @xcite ( @xmath93 ) and @xcite ( @xmath98 ) by using a monte carlo sampling method to estimate a lae lf .
we first draw random galaxies from the uv - continuum lfs . the number of galaxies is defined by the integral of the uv luminosity function at the different redshifts . on the bright end
we integrate to @xmath99 , above which the contribution of galaxies becomes negligible . on the faint end
, we set the integration limit to @xmath100 .
we note , that this is @xmath412 magnitudes below the ly@xmath0 luminosity which is observed at all redshifts . changing @xmath101 above this limit does not change the output of our model .
this faint @xmath101 limit however means extrapolating the observed uv - continuum lfs used from the literature ( usually going down to @xmath102 ) .
so we also verified that the implications of our model are insensitive to changes of the faint end slopes of the uv - continuum lfs and other lf parameters between different studies @xcite . for each of the galaxies drawn from the uv - continuum lf
we then pick a random rest - frame equivalent width from the input distributions and compute the cumulative ly@xmath0 lfs .
we assume no correlation between ew@xmath91 and uv - luminosity for simplicity , although there are hints of less luminous galaxies reaching larger ew@xmath91 compare to more luminous ones ( @xcite but see @xcite and @xcite for a contradictory study ) .
we also assume that every galaxy is emitting ly@xmath0 ( which is then absorbed in the igm and the ew distribution captures the ism physics ) , i.e. , the fraction of ly@xmath0 emission ( @xmath103 ) is 100% for our model .
our models are shown as shaded regions in figure [ fig : model ] , panels a through e. the points show the same weighted averages as in figure [ fig : literature ] and we find that our model is very sensitive to the assumed ew@xmath91 distribution .
this is illustrated by the broad swath of the shaded region indicating the range of values obtained by the @xcite and @xcite ew@xmath91 distributions .
this is not surprising , as from the comparison of the two ew@xmath91 distributions it can be seen that @xcite is missing high ew@xmath91 compared to @xcite which results in a much lower ly@xmath0 lf estimate . in the following
we will assume the @xcite ew@xmath91 distribution as basis because it samples fainter galaxies which contribute to the majority of objects in our sample while @xcite is restricted to uv continuum redshifts and therefore brighter galaxies . to illustrate the dependence on ew@xmath91 further , the dotted , dashed , and dash - dotted lines in figure [ fig : model ] show _ constant _ input rest - frame equivalent widths with ew@xmath104 .
we find good overall agreement between our `` predicted '' lae lf and the observed values up to @xmath15 . but
note in figure [ fig : model ] the observed lae lf moves from the bottom of the predicted range at @xmath105 to the top at @xmath87 .
this indicates the ew@xmath91 distribution appears to be skewing to higher values as found by @xcite ( compare with the lines at constant ew@xmath91 in figure [ fig : model ] ) and is likely caused by decreasing amounts of dust .
in contrast , at @xmath20 the lae lf appears to return to the middle or bottom range of the shaded region predicted by our model . assuming the ( intrinsic ) ew distribution does not change , then a change in the igm is needed to reproduce the observation .
this indicates the igm is becoming more opaque at @xmath20 , suggesting re - ionization finished at @xmath11 . turning to a more qualitative analysis
, we use our model to constrain the change in neutral hydrogen fraction in the igm at @xmath79 . for this , we consider two different possibilities of how we think ly@xmath0 photons get absorbed in the igm .
the two different approaches lead to different imprints of re - ionization in the ly@xmath0 luminosity functions . we consider _
( i ) _ a `` black and white '' process where ly@xmath0 emission of a galaxy is either absorbed or not ( `` patchy / absorption model '' ) and _ ( ii ) _ a smooth process where the ly@xmath0 emission is attenuated by a certain degree ( `` smooth / attenuation model '' ) . the former process will decrease the number of ly@xmath0 emitting galaxies irrespective of their emission strength .
it will lead to a `` global '' shift of the lae lf .
the later process will lower the ly@xmath0 emission in all of the galaxies , preferentially removing galaxies with high ly@xmath0 rest - frame equivalent width .
it will lead to a change in normalization _ and _ shape of the lae lf .
for both of these models we can constrain @xmath8 independently .
we estimate @xmath8 for the former by using the simulations by @xcite , for the later we apply the models by @xcite .
we note that with the current data it is not possible to ( dis)prove one or the other approach .
but we will see that both approaches will lead to the consistent results . in this case
ly@xmath0 is blocked by the neutral igm which results in a decrease of the ly@xmath0 lf for all luminosities .
we tune our model lf to fit the observed lae lfs at @xmath106 5.7 , 6.6 , and 7.7 by adjusting @xmath103 ( the total fraction of galaxies for which ly@xmath0 is not absorbed ) , which is ( at first ) independent of magnitude ( see figure [ fig : reionization ] panels a through c , dotted curves ) .
we find that @xmath103 is almost undistinguishable between @xmath19 but drops by a factor of 4 beyond @xmath107 as it is shown in figure [ fig : reionization ] , panel d by the filled squares .
furthermore , we follow the approach of @xcite and introduce two different values @xmath108 and @xmath109 for simulated galaxies with @xmath110ab and @xmath111ab in order to compare the fraction of ly@xmath0 emitters from our empirical model to real observations at @xmath26 .
this is shown in figure [ fig : reionization ] panels a and b by the dashed line ( we do not apply this split at @xmath30 because of the sparse data ) .
the values for @xmath108 and @xmath109 for ew@xmath112 are shown in panel d ( filled and open circles , respectively ) .
the error bars are estimated by changing the m@xmath113 cut in a range of @xmath114 .
also shown are the observations by @xcite ( light gray ) , @xcite ( black ) , and @xcite ( dark gray ) for galaxies with ew@xmath112 and the same magnitude cut . in general
, we find a good agreement of @xmath115 with the values observed in uv - continuum selected lbgs at @xmath26 .
we find a significant drop of a factor of @xmath116 in the fraction of ly@xmath0 emitters at @xmath30 compared to @xmath6 .
note that the @xcite estimate of @xmath103 for bright galaxies is a factor of @xmath117 higher than the estimates from the other studies .
different selection and sample variance are a very likely cause for this discrepancy . nonetheless , their results support a strong drop of @xmath103 above @xmath107 .
this change in lf can be converted into a neutral hydrogen fraction ( @xmath8 ) by using the results from 186-mpc radiative transfer simulations by @xcite as follows : their figure 4 shows the relative change of the ly@xmath0 lf as a function of neutral hydrogen fraction at @xmath118 assuming full re - ionization at @xmath6 . for example
@xmath119 0.18 , 0.38 , 0.53 , 0.67 , and 0.80 result in a re - scaling of the lf with factors of 0.76 , 0.50 , 0.33 , 0.20 , and 0.05 , respectively .
we then assume that this re - scaling of the lf is directly proportional to the change in the fraction of ly@xmath0 emitters , i.e. , @xmath120 ( see figure [ fig : reionization ] , panel d , blue and red squares ) . assuming the dust extinction properties at @xmath30 are the same as at @xmath6 , we conclude that the drop in @xmath103 implies a neutral hydrogen fraction of _ at least _
@xmath121 at @xmath30 . assuming the dust content of galaxies above @xmath6 is further decreasing and therefore extrapolating @xmath115 from the values at @xmath93 ( see @xcite ) implies even higher limits ( @xmath122 ) .
note that the small change in @xmath103 between @xmath87 and @xmath88 is indicative of little neutral hydrogen .
this is in line with the results by @xcite who suggest the universe is fully ionized at these redshifts .
note that we can estimate @xmath8 without applying our model , by directly taking the ratio of the lae lfs at @xmath87 and @xmath123 and applying again the simulations by @xcite .
this approach leads to consistent results .
having established a lower limit on @xmath8 , we can use the patchy model further to constrain the ly@xmath0 optical depth .
assuming the change in @xmath103 above @xmath6 is due to the igm , it can be associated to the _ average change of ly@xmath0 optical depth @xmath124 under the assumption that re - ionization is completed at @xmath125 _
( i.e. , @xmath126 and @xmath127 ) . note , that this approach is identical to @xcite and we can set @xmath128 , where @xmath129 is defined as in @xcite and @xmath130 by construction . from figure [ fig : reionization ] , panel d we find @xmath129 = [email protected] for @xmath88 ( blue and green squares ) and @xmath129 = [email protected] for @xmath30 ( blue and red squares ) , respectively .
our @xmath88 ( @xmath30 ) value is consistent with the @xmath15 ( @xmath34 ) value of 0.66@xmath132 ( @xmath133 ) found by @xcite @xcite within errors .
we then compute the ly@xmath0 optical depth by equating @xmath134 .
the final result of @xmath135 w.r.t .
@xmath125 is shown in figure [ fig : tau ] .
our limit at @xmath30 is important to constraint @xmath135 as the values at @xmath125 and 7 are almost indistinguishable .
the overall change in optical depth as a function of redshift can be expressed by @xmath136 with @xmath137 .
note that this exponent is a lower limit because of the upper limit in the lae lf at @xmath30 .
we find an increase in optical depth of at least 1.3 between @xmath6 and @xmath34 .
our best fit model is fully consistent with the gunn - peterson optical depth measurements in quasars @xcite , however the functional forms of the estimates lead to different exponents ( see figure [ fig : tau ] ) . in this case
there is no global scaling of the lf as before , however a steepening of the lf may occur because the ew@xmath91 distribution gets skewed to lower ew@xmath91 as the redshift increases beyond @xmath138 ( see also @xcite ) .
we represent the stark et al .
ew distribution in the same manner as @xcite by using a gaussian truncated at negative values .
in contrast to the case outlined before , we now change the width of the ew@xmath91 distribution ( similar to the `` smooth model '' in @xcite ) . as in the case
above , we have to take the difference in evolution between @xmath6 and @xmath24 ( assuming the igm is fully re - ionized at @xmath6 ) .
we therefore start directly with the @xmath6 ew@xmath91 distribution ( see figure [ fig : reionization ] , panel a , dotted curve ) and tune it to fit the @xmath30 limits by changing its width ( dashed - dotted line in figure [ fig : reionization ] , panel c ) . from the final ew@xmath91 distribution at @xmath30 , we compute the cumulative fraction @xmath139 which has now changed w.r.t .
@xmath6 as we have adjusted the width of the ew@xmath91 distribution .
this fractions can be converted into @xmath8 by using the models by @xcite ( using semi - numerical simulations by @xcite ) combining galactic outflow models and large - scale semi - numeric simulations of reionization . from our final ew@xmath91 distribution fitting the limits at @xmath30 we find @xmath140 , @xmath141 , and @xmath142 which translates , by adopting figure 5 in @xcite , into upper limit neutral hydrogen fractions of @xmath8 = @xmath143 , @xmath144 , and @xmath145 , respectively .
note that @xmath8 is more difficult to estimate for smaller ew@xmath91 cuts as @xmath139 approaches unity for all @xmath8 by construction @xcite . taking this into account ,
the limits we find with our second approach are consistent with the results above .
optical depth with redshift with respect to @xmath6 assuming the universe is fully re - ionized by then . under this assumption
, we use the @xcite formalism to find the mean change in ly@xmath0 optical depth with respect to @xmath6 which we assume to be proportional to the change in fraction of ly@xmath0 emitting galaxies .
our limit at @xmath34 is important to constraint @xmath146 which we find to be best fit as @xmath147 , @xmath137 ( solid red line ) .
the strong evolution of at least 1.3 beyond @xmath6 is apparent and could be indicative of a dramatic change in the properties of the igm
. shown along with our best fit is the exponent from the best fit to the evolution of the gunn - peterson optical depth measured on ly@xmath0 , ly@xmath148 , and ly@xmath149 transitions in quasars @xcite.[fig : tau ] ] in summary , we have looked at two different ways how re - ionization can be imprinted in the change of ly@xmath0 lf .
we have considered an absorption model resulting in a global shift of the ly@xmath0 lf and an attenuation model resulting in a skewing of the ew@xmath91 distribution and there for a steepening of the ly@xmath0 lf .
note , that both approaches can fit the observed lae lfs within its uncertainty and we are not able to judge which of the models is right .
however , a skewing of the ew distribution is likely as it seems from the observational data at @xmath87 and @xmath88 that the evolution of the bright end is stronger than at the faint end of the lae lf . in either way
, we are able to constrain @xmath8 using both approaches , resulting in lower limits for the neutral hydrogen fraction between @xmath150 and @xmath151 at @xmath30 .
finally , we stress that our results are based on the assumption that all changes in @xmath103 and the ew@xmath91 distribution are caused by a change in the ionization state of the igm at @xmath79 . however , and
alternative explanation involves an increase of the escape fraction of ionizing photons and would lead to a drop in @xmath103 and thus an overestimation of @xmath8 @xcite . without a changing ionization state of the igm the escape fraction needed to explain the observations is at odds with other studies @xcite .
however , a mixture of changing @xmath8 ( @xmath152 ) and @xmath153 ( @xmath154 ) would be consistent with our results and direct escape fraction measurements .
given these results at @xmath30 it is important to push to higher redshifts to better constrain the evolution of the lae lf . assuming that the lae lf continues to trace the lbg lf at @xmath156
, we can put upper limits on the number of laes that should be found in planned surveys .
the final ultravista nb118 survey @xcite is able to search for potential lae candidates at @xmath155 on 0.9 deg@xmath42 on sky down to @xmath157erg s@xmath29 @xmath158 . assuming this as limiting ly@xmath0 line flux and combined with our model from the lbg uv lf ( optimistically assuming @xmath159 ) it is unlikely that this survey will find laes at this redshift ( expected counts are @xmath160 ) . likewise , with the same assumptions , euclid @xcite is not expected to find laes at @xmath161 with its spectroscopic configuration ( 1.1@xmath162 - @xmath163 , @xmath164erg / s/@xmath40 on 20,000deg@xmath42 ) .
on its proposed deep area ( 40deg@xmath42 ) a flux limit of at least @xmath165erg / s/@xmath40 must be reached to find one lae at @xmath161 .
other space - based spectroscopic surveys like wisps @xcite or 3d - hst @xcite using the hst grism g141 , current flux limits around @xmath166erg s@xmath29 @xmath158 , and area of @xmath167 arcmin@xmath42 need to be substantially ( roughly 5 times ) deeper to find laes at @xmath155 .
very deep small area blind imaging surveys with instruments on 8 - 10 m telescopes such as hawk - i ( @xmath168 ) or mosfire ( @xmath169 ) must reaching flux limits of @xmath170erg s@xmath29 @xmath158 in nb118 to pick up one lae at @xmath155 on a total of @xmath171 pointings .
we have presented follow - up observations on two bright lae candidates at @xmath30 using mosfire .
we rule out any line emission at a level of several @xmath72 for both objects .
the limits inferred from these non - detections suggest a strong evolution of the lae lf between @xmath90 , consistent with what is seen in lbg samples .
we create an empirical model using the observed lbg uv continuum lfs and ly@xmath0 rest - frame equivalent width distributions to understand the interplay between lae and uv continuum selected galaxies .
we find that our model and the observed lae lf follow each other , but note a secondary effect which is due to a change in the ew@xmath91 distribution of the galaxies as a function of redshift . from this differential evolution and assuming two different models on ly@xmath0 absorption , we find consistent lower limits on the neutral hydrogen fraction at @xmath30 of 50 - 70% .
furthermore , we find a strong evolution in the ly@xmath0 optical depth at @xmath79 which can be characterized by @xmath5 .
all in all , our results are indicative of a continuation of strong evolution in the igm beyond @xmath107 .
we would like to acknowledge the support of the keck observatory staff who made these observations possible as well as b. trakhtenbrot and w. hartley for valuable discussions .
we also thank nick konidaris for providing and supporting the mosfire reduction pipeline and gwen rudie for providing the mosfire exposure time calculator .
af acknowledges support from the swiss national science foundation .
af also thanks caltech for hospitality while this article was worked on .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain . , r. g. , et al .
2003 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 4841 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . m. iye & a. f. m. moorwood , 525539 , s. , rhoads , j. , dey , a. , stern , d. , & spinrad , h. 2001 , in astronomical society of the pacific conference series , vol .
240 , gas and galaxy evolution , ed .
j. e. hibbard , m. rupen , & j. h. van gorkom , 97 ccccccc 3.1 & 356 & 41/- & 12% & @xmath172 ( nb503 ) & @xmath173&@xcite + 3.1 & 160 & - & 0% & @xmath174 ( nb5000 ) & @xmath175 & @xcite + 3.7 & 101 & 26/- & 26% & @xmath176 ( nb570 ) & @xmath173&@xcite + 4.9 & 87 & - & 0% & @xmath177 ( nb711 ) & @xmath178&@xcite + cccccccc 5.7 & 89 & 46/66 + 8 & 55% & @xmath177 ( nb816 ) & 5@xmath72 , 2@xmath63 aperture & @xmath179&@xcite + 5.7 & @xmath180 & 87/140 & 100% & @xmath172 ( nb816 ) & 5@xmath72 , 3@xmath63 aperture & @xmath181&@xcite + 5.7 & 401 & 17/29 & 4% & @xmath177 ( nb816 ) & 5@xmath72 , 2@xmath63 aperture & @xmath173&@xcite + 5.7 & 119 & - & 0% & @xmath182 ( nb816 ) & 5@xmath72 , 2@xmath63 aperture & @xmath183 & @xcite + 5.7 & 89 & 28/39 + 6/24 & 36% & @xmath184 ( nb816 ) & 3@xmath72 , 2@xmath63 aperture & @xmath185&@xcite + 5.7 & 56@xmath186 & 30/35 & 55% & - & & @xmath187&@xcite + 6.6 & 207 ( + 58 ) & 16/24 ( + 16.22 + 1 ) & 13% & @xmath188 ( nb921 ) & 3@xmath72 , 2@xmath63 aperture & @xmath173&@xcite + 6.6 & 58 & 42/52 + 3 & 74% & @xmath177 ( nb921 ) & 5@xmath72 , 2@xmath63 aperture & @xmath179&@xcite + 6.6 & @xmath189 & 30/70 & 100% & @xmath190 ( nb912 ) & 5@xmath72 , 3@xmath63 aperture & @xmath181 & @xcite + 6.6 & 61 & 12/23 & 20% & - & & @xmath191&@xcite + 7.7 & 4 & 0/2 & 0% & @xmath192 ( unb1056 ) & 50% compl . ,
auto aper . &
@xmath185 & @xcite + 7.7 & 0 & - & - & @xmath177 ( nb1060 ) & 5@xmath72 , @xmath193aperture & @xmath194 & @xcite + 7.7 & 7 & 0/5 & 0% & @xmath190 ( nb1060 ) & 4@xmath72 , 1.5apert .
(= 50% compl . )
& @xmath195 & @xcite + 7.7 & 4 & - & 0% & @xmath196 ( unb1063 ) & 50% completenes & @xmath185 & @xcite + | we present spectroscopic follow - up observations on two bright ly@xmath0 emitter ( lae ) candidates originally found by @xcite at a redshift of @xmath1 using the multi - object spectrometer for infra - red exploration ( mosfire ) at keck .
we rule out any line emission at the @xmath2 level for both objects , putting on solid ground a previous null result for one of the objects .
the limits inferred from the non - detections rule out the previous claim of no or even reversed evolution between @xmath3 in the ly@xmath0 luminosity function ( lf ) and suggest a drop in the ly@xmath0 luminosity function consistent with that seen in lyman break galaxy ( lbg ) samples .
we model the redshift evolution of the lae lf using the lbg uv continuum lf and the observed rest - frame equivalent width distribution . from the comparison of our empirical model with the observed lae distribution
, we estimate lower limits of the neutral hydrogen fraction to be 50 - 70% at @xmath4 . together with this
, we find a strong evolution in the ly@xmath0 optical depth characterized by @xmath5 beyond @xmath6 indicative of a strong evolution of the igm .
finally , we extrapolate the lae lf to @xmath7 using our model and show that it is unlikely that large area surveys like ultravista or euclid pick up laes at this redshift assuming the current depths and area . |
the origin of gravity and the laws of newton is a long standing issue . even after the discovery of general relativity by einstein ,
the answer about the origin of universal gravity is still far from satisfying because of two reasons : ( i ) there is still serious problem in the unification of quantum mechanics and general relativity ; ( ii ) it is still not clear how to establish a unified theory of electromagnetic , weak , strong and gravitational forces . in a recent work ,
verlinde @xcite emphasized again the end of gravity as a fundamental force , which has already been suggested by jacobson @xcite almost fifteen years ago .
a related idea was also proposed by padmanabhan @xcite .
verlinde s work has ignited intensive studies on the thermodynamical origin of gravity .
these advances were motivated by the fact that the gravity law closely resembles the laws of thermodynamics and hydrodynamics bekenstein , bardeen , hawking , davies , unruh .
following verlinde s work @xcite , we reconsider the assumption of the change of entropy @xmath0 for an object having a displacement @xmath1 and the unruh effect which assuming a temperature for an accelerated object .
this assumption and unruh effect play an essential role in verlinde s derivations on the universal gravity and the laws of newton . in the present work ,
simple derivations are given for this assumption and unruh effect with the standard theory about quantum mechanics , special relativity and vacuum background based on the gauge field theory .
our derivations establish further the logic basis of regarding the gravity as an entropic force .
we also stress the local thermal equilibrium @xcite in deriving the universal gravity .
in addition to the dimensional consideration to get the planck length , we try to argue that the planck length is the coherence length of the fluctuating vacuum background in thermal equilibrium . as for the equivalence principle ,
the derivation of the newton s law of gravity shows clearly that the gravitational mass is the same physical concept as the inertial mass .
therefore , general relativity is the natural generalization of the special relativity with the inclusion of gravity , where the weak equivalence principle is not an assumption any more .
in the original work by verlinde @xcite , the following postulation motivated by bekenstein s work @xcite about black holes and entropy , plays a key role in deriving the newton s law of gravity@xmath2 in the above formula , @xmath0 is the change of entropy after a displacement @xmath3 for a particle with mass @xmath4 .
here we reconsider the physical mechanism of this postulation based on quantum mechanics , special relativity and vacuum background . from the special relativity , we have@xmath5from the quantum mechanics , the eigenfrequency is then@xmath6if the space is regarded as a sort of fluctuating vacuum background due to the existence of various gauge fields
, there is a strong coupling between the matter and the vacuum background . as a consequence , from the view of the vacuum background
, we have the following coherence length for the particle:@xmath7the factor @xmath8 in the above equation is due to the fact that the gauge field in the vacuum has just the propagation velocity of @xmath8 .
if @xmath9 , we have@xmath10when the coupling between the particle and the vacuum background is considered , the coherence length @xmath11 gives the location resolution of the particle from the view of the vacuum background . in fig .
1(a ) , we show the motion of a particle along the dashed line .
because of the location resolution from the view of the vacuum background , the dashed line is partitioned by the box with side length @xmath12 . at time @xmath13 , the wavepacket of the particle is shown by the dashed line . at a later time
@xmath14 ( @xmath15 ) , the wavepacket of the particle is shown by the solid line . at time @xmath14 , the information of the location of the particle at time @xmath16 recorded by the vacuumbackground will be lost because of two reasons : ( 1 ) at time @xmath14 , although the location of the particle with spatial resolution @xmath12 is recorded by the vacuum background , the velocity information is highly uncertain . from @xmath17 ,
the velocity uncertainty of the particle is @xmath18 . in this situation , at time @xmath14 , after the position of the particle is recorded by the vacuum background , the history of the particle is lost from the view of the vacuum background .
( 2 ) at time @xmath13 , although the location with spatial resolution @xmath12 is recorded by the vacuum background , it will be lost rapidly after a displacement at time @xmath14 . at time @xmath13 , the location record is due to the coupling between the particle and the vacuum background . because the vacuum background is highly fluctuating , at time @xmath14 , the location information at time @xmath13 will be lost because of the re - establishment of the thermal equilibrium in the vacuum background . ) .
( c ) shows the essential difference between @xmath12 and @xmath19 . ]
the above analysis shows that there is an entropy increase of @xmath20 after the displacement of @xmath12 . from this result , it is straightforward to get@xmath21with @xmath3 being the overall length of the particle trajectory , irrelevant to the trajectory shape . by setting @xmath22 , we get eq .
( [ entropy ] ) . as shown in fig .
1(b ) , for a trajectory with length @xmath23 , the bits of information @xmath24 .
however , from the view of vacuum background , all these information are lost at the final location of this trajectory .
the linear relation between @xmath0 and @xmath3 is due to the fact that the overall information in units of bit is proportional to the overall length of a trajectory . to justify further the above derivation
, we give here several discussions .
\(1 ) in the above derivation , we assume a strong coupling between the particle and the vacuum background , so that the location of the particle can be recorded by the vacuum background . as shown in sec .
iv , the temperature of the vacuum background is about @xmath25 .
thus , the vacuum background is in fact full of various gauge fields , virtual matter - antimatter pairs .
this implies a strong coupling between the particle and the vacuumbackground .
\(2 ) one should distinguish two sorts of coherence lengths : the coherence length @xmath12 given by eq .
( [ coherencelength ] ) and the ordinary thermal de broglie wave length . in the non - relativistic approximation ,
the thermal de broglie wave length reflects the spatial coherence length which is @xmath26 .
we stress that @xmath12 and @xmath19 have different physical origins .
@xmath12 originates from the coupling between the particle and the vacuum background .
because of the extremely high temperature of the vacuum background and strong coupling , the special relativity and quantum mechanics are used to calculate @xmath12 .
in contrast to @xmath12 , @xmath19 originates from the coupling or interaction between the particle we studied and other particles or environment . to clearly elaborate this
, we take a hydrogen atom as an example , which consisting of an electron and a proton . for a system consisting of a large number of hydrogen atoms , @xmath27 with @xmath28 being the ordinary temperature in thermal equilibrium .
@xmath19 physically originates from the interatomic interaction or the coupling with the environment . for bose - einstein condensation of atomic hydrogen @xcite
, @xmath19 can arrive at @xmath29 .
for the proton in a hydrogen , the coherence length of the proton is about the diameter of the hydrogen atom , _
i. e. _ @xmath30 .
this sort of @xmath31 can be obtained through the non - relativistic approximation .
it is clear that @xmath31 physically originates from the electromagnetic interaction between proton and electron in a same hydrogen . for the quark in the proton ,
the coherence length @xmath32 of the quark is about the radius of the proton ( which is about @xmath33 ) . because the proton is composed of three quarks , we approximate the quark mass ( rather than the naked quarks ) as one third of the proton mass . from eq .
( [ coherencelength ] ) , we have @xmath34 @xmath35 for the quark . quite interesting , @xmath32 agrees with this value of @xmath12 .
this is understandable , because the quark in the proton has strong coupling with the vaccum background due to strong interaction .
this means that the coherence length of quarks should be always of the order of several @xmath36 . for a free quark
, its velocity uncertainty could be smaller than @xmath8 in principle , and thus larger than @xmath12 .
therefore , the existence of a free quark contradicts with the coherence length of @xmath12 .
this implies in a simple way the well - known quark confinement .
( 3 ) because @xmath19 , @xmath31 and @xmath32 have different physical origin , they have different physical significance .
in particular , for the situations of @xmath37 and @xmath38 , @xmath19 still has physical significance for a whole atom . for sufficient low temperature
, there is still bose - einstein condensation of dilute atomic gas , regarded of much smaller @xmath31 and @xmath32 .
the physical significance of @xmath39 is due to the fact that when interatomic interaction is addressed , the freedom of internal motion of proton and electron in an atom is frozen out .
it is similar for @xmath31 , because the quark confinement means that the interaction between quark and vacuum background is screened . in a sense
, the strong coupling between the quark and vacuumbackground is blurred when the coherence effect of the whole atom or proton is addressed . in fig .
1(c ) , we show @xmath40 ( @xmath41 ) and @xmath19 for an atom
. nevertheless , we will show in due course that the presence of @xmath11 and thus eq . ( [ entropy ] ) will play important role in the derivation of gravity .
\(4 ) because of the blurred effect for @xmath11 ( @xmath41 ) , one expects that the gravity between two atoms is negligible , compared with other interactions . for example , it is clear that the formation of bose - einstein condensation of dilute atomic gases is completely irrelevant to the gravity interaction between atoms . because the entropy is an extensive quantity , eq .
( [ entropy ] ) can be directly generalized to an object comprising of a large number of fundamental particles . in this situation
, @xmath0 can be very large for sufficiently large mass of an object , and the gravity between different objects will play important role .
the unruh temperature @xcite is the effective temperature experienced by a uniformly accelerating object in a vacuum field . here
we give a simple derivation of unruh effect from eq .
( [ entropy ] ) . considering a particle with acceleration @xmath42
, we have @xmath43 . here
@xmath44 can be also regarded as the relative velocity of a reference system . from eq .
( [ entropyalpha ] ) , we have @xmath45 from eq .
( [ energy ] ) , we have@xmath46using @xmath47 , we get@xmath48this gives the famous unruh formula by setting @xmath22 . from the formula@xmath49we have@xmath50using eqs .
( [ entropyalpha ] ) and ( [ unruh ] ) , it is easy to get the following inertia law@xmath51
following the verlinde s idea , we give a derivation of newton s law of gravitation . in our derivation
, we will stress the local thermal equilibrium and the coherence length of the fluctuating vacuum background .
we consider the emergent force between two objects with mass @xmath52 and @xmath4 , shown in fig .
2(a ) . to solve this problem , we first consider the situation shown in fig .
2(b ) , where the mass @xmath53 is uniformly distributed in a region between two spherical surfaces with radius @xmath54 and @xmath55 . assuming @xmath56 , the overall volume of this region is @xmath57 .
this region is divided into @xmath58 equal parts . in this situation , in every part , the mass is @xmath59 .
2(c ) gives a cross section diagram of fig .
we consider the emergent force with the vacuum background as the medium .
more precisely , both objects have strong coupling with the vacuum background , and local thermal equilibrium will result in an emergent force .
because the coupling with the vacuumbackground is considered , special relativity should be used from the beginning . from the mass - energy relation @xmath60
, we get an effective temperature @xmath61 , determined by@xmath62for a particle with @xmath4 approaching this sphere , the unruh temperature @xmath63 of this particle should be equal to @xmath61 because of thermal equilibrium , i.e.@xmath64by using the unruh formula ( [ unruh ] ) , we get @xmath65by introducing the gravitation constant @xmath66 , we get @xmath67 , which is just the planck length
. we will discuss the origin of this planck length in due course . in the above derivation
, the unruh formula is used , while eq .
( [ entropy ] ) formula is not directly used .
there is another derivation of eq .
( [ ga ] ) by using eq .
( [ entropy ] ) . for the particle @xmath4 , we have@xmath68using eq . ( [ thermalequilibrium ] ) , we get@xmath69using further eq . ( [ entropy ] ) , we get@xmath70from eq .
( [ ga ] ) ( derived using unruh s formula ) and ( [ gf ] ) ( derived using eq . ( [ entropy ] ) ) , we get the result of the inertia law @xmath71 .
this is natural , because the unruh formula can be derived from eq .
( entropy ) and thermodynamical relation , while the inertia law can be derived from eq .
( [ entropy ] ) , thermodynamical relation and unruh formula . based on the consideration of central symmetry , we get further@xmath72 compared with verlinde s work , the merit of the above derivation lies in that the role of local thermal equilibrium is stressed in the derivation of the newton s law of gravitation .
now we turn to consider an object of mass @xmath73 locating at the position in fig .
2 ( a ) . from gauss s flux theorem and eq .
( [ gfv ] ) , we have@xmath74this shows that the gravitation is still given by eq .
( [ gfv ] ) for this situation . for a particle with mass @xmath52 , from @xmath75 for another particle with mass @xmath4 and the unruh formula
, there is an effective vacuum temperature due to the presence of @xmath52.@xmath76this leads to a physical picture that the presence of @xmath52 establishes an effective temperature field distribution .
we assume that the object with mass @xmath52 is uniformly distributed in the sphere with radius @xmath77 .
( [ mtemperature ] ) can be used to calculate @xmath61 for @xmath78 . for @xmath79 , however , using eq .
( gauss ) and unruh formula , the temperature is@xmath80the overall temperature distribution is shown in fig .
3 . therefore , the maximum temperature is@xmath81from the mass - energy relation , even all the energy is transferred into temperature with only one freedom
, we get a limit temperature@xmath82obvious , there is a request of@xmath83this means that @xmath84this result shows that any object having strong coupling with the vacuum background can not be distributed within a sphere of radius @xmath85 , irrelevant to its mass .
this provides a physical mechanism why in the previous calculation of @xmath61 based on eq .
( [ tm ] ) , the smallest region for the distribution of @xmath52 is @xmath86 . .
] we stress that even the vacuumbackground is completely continuous and smooth , there is still the coarse distribution of @xmath52 due to gravitation , which originates from the coupling between matter and the vacuumbackground
. nevertheless , the above analysis gives a possible physical mechanism for the coarse graining of the space with the planck length @xmath87 .
here we consider further the physical significance of the planck length @xmath88 .
the coarse graining of the vacuum background with length @xmath87 means that the vacuum energy in the smallest space unit is@xmath89by using the mass - energy relation , the effective mass in this space unit is@xmath90the coherence length of the vacuum background is then@xmath91we see that @xmath87 gives also the coherence length of the vacuum fluctuations .
this implies strongly again the coarse graining of space with the planck length @xmath87 .
one may argue that if there are vacuumfluctuations with a finite coherence length @xmath12 , the increasing of @xmath92 will lower the overall energy of the vacuum background , and thus is energy stable for infinite @xmath12 .
based on the understanding of gauge theory , however , it is well - known that @xmath12 should be finite and the vacuum energy density should be nonzero .
therefore , a more natural question is what length of @xmath87 will make the vacuum background dynamically and thermodynamically stable . as shown above , the gravitation is no longer a fundamental force .
it physically originates from the coupling between matter and vacuum background through gauge field .
when the following expression is satisfied@xmath93it is understood that the vacuumbackground becomes stable .
this gives@xmath94using further @xmath95 , we get again @xmath96 . from the above results , the temperature of the vacuum background is@xmath97 in this situation
, we have @xmath98 . as emphasized above
, this extremely high temperature is the thermal equilibrium due to the finite coherence length of highly fluctuating vacuum background and the gravity between these fluctuating vacuum background .
this extremely high temperature of the vacuum background also means that the assumption of strong coupling between an object and the vacuum background in deriving the universal gravity is reasonable .
it is well known that the equivalence principle is based on the assumption that the gravitational mass @xmath99 is equal to the inertial mass @xmath100 .
it is still a mystery why @xmath101 because it seems that the gravitational mass and inertial mass are completely different physical concept , if the universal gravitation is still regarded as a fundamental force .
the above analysis has shown that newton s law of gravitation can be derived from the special relativity where the symbols @xmath52 and @xmath4 can both be regarded as the mass in the mass - energy relation . in all our derivations , we do not need especially introduce the gravitational mass at all . in this situation ,
the gravitational mass and inertial mass are in fact the same physical concept .
hence , the weak equivalence principle is also a derivable law from special relativity , quantum mechanics and true fundamental forces .
the significance of excluding the gravity as a fundamental force has been pointed out in refs . @xcite . in fig .
4 , we give the logical basis and the basic clues of our work to get newton s law of gravity , inertia law and weak equivalence principle .
all our derivations depend on the special relativity , quantum mechanics and vacuum background .
the information theory or thermodynamics is a tool to analysis the complex many - body system , similarly to the studies for a classical system with thermodynamics .
compared with verlinde s work , in the present work , the concept of holographic screen is not absolutely necessary to derive the universal gravity and the laws of newton . | following verlinde s recent work on the origin of gravity and the laws of newton , we consider further the origin of the starting point of verlinde s work and unruh effect .
simple derivations are given for this starting point and unruh effect with the standard theory about quantum mechanics , special relativity and vacuum background based on the gauge field theory .
our studies give the physical origin of the planck length .
the present work supports further the idea that gravity is not a fundamental force . in the present work ,
the concept of holographic screen is not absolutely necessary to derive the universal gravity and the laws of newton . |
polynomial interpretations are a simple yet useful technique for proving termination of term rewrite systems ( trss , for short ) .
while originally conceived in the late seventies by lankford @xcite as a means for establishing direct termination proofs , polynomial interpretations are nowadays often used in the context of the dependency pair ( dp ) framework @xcite . in the classical approach of lankford ,
one considers polynomials with integer coefficients inducing polynomial algebras over the well - founded domain of the natural numbers . to be precise
, every @xmath0-ary function symbol @xmath1 is interpreted by a polynomial @xmath2 in @xmath0 indeterminates with integer coefficients , which induces a mapping or _ interpretation _ from terms to integer numbers in the obvious way . in order to conclude termination of a given trs
, three conditions have to be satisfied .
first , every polynomial must be _ well - defined _
, i.e. , it must induce a well - defined polynomial function @xmath3 over the natural numbers .
in addition , the interpretation functions @xmath4 are required to be _ strictly monotone _ in all arguments . finally , one has to show _ compatibility _ of the interpretation with the given trs .
more precisely , for every rewrite rule @xmath5 , the polynomial @xmath6 associated with the left - hand side must be greater than @xmath7 , the corresponding polynomial of the right - hand side , i.e. , @xmath8 for all values of the indeterminates . already back in the seventies , an alternative approach using polynomials with real coefficients instead of integers was proposed by dershowitz @xcite .
however , as the real numbers @xmath9 equipped with the standard order @xmath10 are not well - founded , a subterm property is explicitly required to ensure well - foundedness .
it was not until 2005 that this limitation was overcome , when lucas @xcite presented a framework for proving polynomial termination over the real numbers , where well - foundedness is basically achieved by replacing @xmath10 with a new ordering @xmath11 requiring comparisons between terms to not be below a given positive real number @xmath12 .
moreover , this framework also facilitates polynomial interpretations over the rational numbers .
thus , one can distinguish three variants of polynomial interpretations , polynomial interpretations with real , rational and integer coefficients , and the obvious question is : what is their relationship with regard to termination proving power ? for knuth - bendix orders it is known @xcite that extending the range of the underlying weight function from natural numbers to non - negative reals does not result in an increase in termination proving power . in 2006
lucas @xcite proved that there are trss that can be shown polynomially terminating by polynomial interpretations with rational coefficients , but can not be shown polynomially terminating using polynomials with integer coefficients only .
likewise , he proved that there are trss that can be handled by polynomial interpretations with real ( algebraic ) coefficients , but can not be handled by polynomial interpretations with rational coefficients . in this article
we extend these results and give a complete comparison between the various notions of polynomial termination .. ] in general , the situation turns out to be as depicted in figure [ fig : summary ] , which illustrates both our results and the earlier results of lucas @xcite .
( 0,0 ) node[anchor = south west ] terminating trss rectangle ( 13cm,4 cm ) ; ( 7cm,20 mm ) ellipse ( 36 mm and 13 mm ) ; ( 86mm,25 mm ) ellipse ( 9 mm and 9 mm ) ; ( 75mm,20 mm ) ellipse ( 2 cm and 10 mm ) ; ( 7cm,20 mm ) ellipse ( 36 mm and 13 mm ) ; ( 32mm,12 mm ) rectangle + + ( 7mm,14 mm ) ; ( 36.5mm,20 mm ) node @xmath13 ; ( 52mm,17.5 mm ) rectangle + + ( 5mm,5 mm ) ; ( 54.5mm,20 mm ) node @xmath14 ; ( 92mm,27 mm ) ellipse ( 11 mm and 11 mm ) ; ( 79mm,22 mm ) rectangle + + ( 5mm,5 mm ) ; ( 81.5mm,24.5 mm ) node @xmath15 ; ( 45mm,20 mm ) node @xmath16 ; ( 70mm,20 mm ) node @xmath17 ; ( 98mm,24 mm ) node @xmath18 ; ( 95mm,34 mm ) node @xmath19 ; in particular , we prove that polynomial interpretations with real coefficients subsume polynomial interpretations with rational coefficients .
moreover , we show that polynomial interpretations with real or rational coefficients do not subsume polynomial interpretations with integer coefficients by exhibiting the trs @xmath20 in section [ sect : nvsr ] .
likewise , we prove that there are trss that can be shown terminating by polynomial interpretations with real coefficients as well as by polynomial interpretations with integer coefficients , but can not be shown terminating using polynomials with rational coefficients only , by exhibiting the trs @xmath21 in section [ sect : nrvsq ] .
the trss @xmath22 and @xmath23 can be found in section [ sect : incrementality ] .
the remainder of this article is organized as follows . in section [
sect : prelim ] , we introduce some preliminary definitions and terminology concerning polynomials and polynomial interpretations . in section [
sect : qvsr ] , we show that polynomial interpretations with real coefficients subsume polynomial interpretations with rational coefficients .
we further show that for polynomial interpretations over the reals , it suffices to consider real algebraic numbers as interpretation domain .
section [ sect : nvsr ] is dedicated to showing that polynomial interpretations with real or rational coefficients do not subsume polynomial interpretations with integer coefficients .
then , in section [ sect : nrvsq ] , we present a trs that can be handled by a polynomial interpretation with real coefficients as well as by a polynomial interpretation with integer coefficients , but can not be handled using polynomials with rational coefficients . in section [
sect : incrementality ] , we show that the relationships in figure [ fig : summary ] remain true if incremental termination proofs with polynomial interpretations are considered .
we conclude in section [ sect : conclusion ] .
this paper is an extended version of @xcite , which contained the result of section [ sect : nvsr ] .
the results in sections [ sect : qvsr ] , [ sect : nrvsq ] and [ sect : incrementality ] are new .
as usual , we denote by @xmath15 , @xmath24 , @xmath14 and @xmath9 the sets of natural , integer , rational and real numbers , respectively .
an _ irrational _ number is a real number , which is not in @xmath14 . given some @xmath25 and @xmath26 , @xmath27 denotes the standard order of the respective domain and @xmath28 .
a sequence of real numbers @xmath29 _ converges _ to the _ limit _
@xmath30 if for every real number @xmath31 there exists a natural number @xmath32 such that the absolute distance @xmath33 is less than @xmath34 for all @xmath35 ; we denote this by @xmath36 . as convergence in @xmath37
is equivalent to componentwise convergence , we use the same notation also for limits of converging sequences of vectors of real numbers @xmath38 .
a real function @xmath39 is _ continuous _ in @xmath37 if for every converging sequence @xmath38 it holds that @xmath40 . finally , as @xmath14 is dense in @xmath9 , every real number is a rational number or the limit of a converging sequence of rational numbers . for any ring @xmath41 ( e.g. @xmath24 , @xmath14 , @xmath9 ) ,
we denote the associated _ polynomial ring _ in @xmath0 _ indeterminates _ @xmath42 by @xmath43 $ ] , the elements of which are finite sums of products of the form @xmath44 , where the _ coefficient _ @xmath45 is an element of @xmath41 and the exponents @xmath46 in the _ monomial _ @xmath47 are non - negative integers .
if @xmath48 , we call a product @xmath44 a _ term_. the _ degree _ of a monomial is just the sum of its exponents , and the degree of a term is the degree of its monomial .
an element @xmath49 $ ] is called an _ ( @xmath0-variate )
polynomial with coefficients in @xmath41_. for example , the polynomial @xmath50 is an element of @xmath51 $ ] , the ring of all univariate polynomials with integer coefficients . in the special case @xmath52 , a polynomial @xmath53
$ ] can be written as follows : @xmath54 ( @xmath55 ) .
for the largest @xmath56 such that @xmath57 , we call @xmath58 the _ leading term _ of @xmath59 , @xmath60 its _ leading coefficient _ and @xmath56 its _ degree _ , which we denote by @xmath61 . a polynomial @xmath53 $ ]
is said to be _ linear _ if @xmath62 , and _ quadratic _ if @xmath63 .
we assume familiarity with the basics of term rewriting and polynomial interpretations ( e.g. @xcite ) .
the key concept for establishing ( direct ) termination of trss via polynomial interpretations is the notion of well - founded monotone algebras as they induce reduction orders on terms .
[ def : ma ] let @xmath64 be a signature , i.e. , a set of function symbols equipped with fixed arities .
an @xmath64-_algebra _ @xmath65 consists if a non - empty _ carrier _ set @xmath66 and a collection of interpretation functions @xmath67 for each @xmath0-ary function symbol @xmath68 .
the _ evaluation _ or _ interpretation _
@xmath69_\mathcal{a}(t)$ ] of a term @xmath70 with respect to a variable assignment @xmath71 is inductively defined as follows : @xmath72_\mathcal{a}(t ) = \begin{cases } \alpha(t ) & \text{if $ t \in \mathcal{f}$ } \\
f_a([\alpha]_\mathcal{a}(t_1),\dots,[\alpha]_\mathcal{a}(t_n ) ) & \text{if $ t = f({{t_1},\dots,{t_{n}}})$ } \end{cases}\ ] ] let @xmath73 be a binary relation on @xmath66
. for @xmath74 , an interpretation function @xmath75 is _ monotone in its @xmath76-th argument _ with respect to @xmath73 if @xmath77 implies @xmath78 for all @xmath79 .
it is said to be _ monotone _ with respect
to @xmath73 if it is monotone in all its arguments .
we define @xmath80 as @xmath69_\mathcal{a}(s ) \sqsupset [ \alpha]_\mathcal{a}(t)$ ] for all assignments @xmath81 . in order to pave the way for incremental polynomial termination in section [ sect : incrementality ] , the following definition is more general than what is needed for direct termination proofs .
[ def : wma ] let @xmath82 be an @xmath64-algebra together with two binary relations @xmath83 and @xmath84 on @xmath66 .
we say that @xmath85 and a trs @xmath86 are _
( weakly ) compatible _ if @xmath87 ( @xmath88 ) for each rewrite rule @xmath89 .
an interpretation function @xmath90 is called strictly ( weakly ) monotone if it is monotone with respect to @xmath83 ( @xmath84 ) .
the triple @xmath82 ( or just @xmath65 if @xmath83 and @xmath84 are clear from the context ) is a _ weakly ( strictly ) monotone @xmath64-algebra _
if @xmath83 is well - founded , @xmath91 and for each @xmath68 , @xmath90 is _ weakly ( strictly ) monotone_. it is said to be an _ extended monotone @xmath64-algebra _ if it is both weakly monotone and strictly monotone .
finally , we call @xmath82 a _ well - founded monotone @xmath65-algebra _ if @xmath83 is a well - founded order on @xmath66 , @xmath84 is its reflexive closure , and each interpretation function is strictly monotone .
it is well - known that well - founded monotone algebras provide a complete characterization of termination .
a trs is terminating if and only if it is compatible with a well - founded monotone algebra .
[ def : polyint_n ] a _ polynomial interpretation over @xmath15 _ for a signature @xmath64 consists of a polynomial @xmath92 $ ] for every @xmath0-ary function symbol @xmath68 such that for all @xmath68 the following two properties are satisfied : 1 .
_ well - definedness _
: @xmath93 for all @xmath94 , 2 .
_ strict monotonicity _ of @xmath4 in all arguments with respect to @xmath95 , the standard order on @xmath15 . due to well - definedness ,
each of the polynomials @xmath4 induces a function from @xmath96 to @xmath15 .
hence , the pair @xmath97 constitutes an @xmath64-algebra over the carrier @xmath15 .
now @xmath98 where @xmath99 is the reflexive closure of @xmath95 constitutes a well - founded monotone algebra , and we say that a polynomial interpretation over @xmath15 is _ compatible _ with a trs @xmath86 if the well - founded monotone algebra @xmath98 is compatible with @xmath86 . finally , a trs is _ polynomially terminating over @xmath15 _ if it admits a compatible polynomial interpretation over @xmath15 . in the sequel
, we often identify a polynomial interpretation with its associated @xmath64-algebra .
[ rem : isom ] in principle , one could take any set @xmath100 ( or even @xmath101 ) instead of @xmath15 as the carrier for polynomial interpretations
. however , it is well - known @xcite that all these sets are order - isomorphic to @xmath15 and hence do not change the class of polynomially terminating trss . in other words ,
a trs @xmath86 is polynomially terminating over @xmath15 if and only if it is polynomially terminating over @xmath100 .
thus , we can restrict to @xmath15 as carrier without loss of generality .
the following simple criterion for strict monotonicity of a univariate quadratic polynomial will be used in sections [ sect : nvsr ] and [ sect : nrvsq ] .
[ lem : nquadmon ] the quadratic polynomial @xmath102 with @xmath103 is strictly monotone and well - defined if and only if @xmath104 , @xmath105 , and @xmath106 .
now if one wants to extend the notion of polynomial interpretations to the rational or real numbers , the main problem one is confronted with is the non - well - foundedness of these domains with respect to the standard orders @xmath107 and @xmath10 . in @xcite ,
this problem is overcome by replacing these orders with new non - total orders @xmath11 and @xmath108 , the first of which is defined as follows : given some fixed positive real number @xmath12 , @xmath109 analogously , one defines @xmath108 on @xmath14 .
thus , @xmath11 ( @xmath108 ) is well - founded on subsets of @xmath9 ( @xmath14 ) that are bounded from below .
therefore , any set @xmath110 ( @xmath111 ) could be used as carrier for polynomial interpretations over @xmath9 ( @xmath14 ) . however , without loss of generality we may restrict to @xmath112 ( @xmath113 ) because the main argument of remark [ rem : isom ] also applies to polynomials over @xmath9 ( @xmath14 ) , as is already mentioned in @xcite .
[ def : polyint_r ] a _ polynomial interpretation over @xmath9 _ for a signature @xmath64 consists of a polynomial @xmath114 $ ] for every @xmath0-ary function symbol @xmath68 and some positive real number @xmath115 such that @xmath116 is well - defined over @xmath112 , i.e. , @xmath117 for all @xmath118 .
analogously , one defines polynomial interpretations over @xmath14 by the obvious adaptation of the definition above .
let @xmath119 .
as for polynomial interpretations over @xmath15 , the pair @xmath120 constitutes an @xmath64-algebra over the carrier @xmath121 due to the well - definedness of all interpretation functions .
together with @xmath122 and @xmath123 , the restrictions of @xmath124 and @xmath125 to @xmath121 , we obtain an algebra @xmath126 , where @xmath122 is well - founded ( on @xmath121 ) and @xmath127 .
hence , if for each @xmath68 , @xmath128 is weakly ( strictly ) monotone , that is , monotone with respect to @xmath123 ( @xmath122 ) , then @xmath126 is a weakly ( strictly ) monotone @xmath64-algebra .
however , unlike for polynomial interpretations over @xmath15 , strict monotonicity of @xmath126 does not entail weak monotonicity as it can very well be the case that an interpretation function is monotone with respect to @xmath122 but not with respect to @xmath123 .
let @xmath119 .
a polynomial interpretation over @xmath129 is said to be _ weakly ( strictly ) monotone _ if the algebra @xmath126 is weakly ( strictly ) monotone .
similarly , we say that a polynomial interpretation over @xmath129 is _
( weakly ) compatible _ with a trs @xmath86 if the algebra @xmath126 is ( weakly ) compatible with @xmath86 .
finally , a trs @xmath86 is _ polynomially terminating over @xmath129 _ if there exists a polynomial interpretation over @xmath129 that is both compatible with @xmath86 and strictly monotone .
we conclude this section with a more useful characterization of monotonicity with respect to the orders @xmath130 and @xmath131 than the one obtained by specializing definition [ def : wma ] . to this end
, we note that a function @xmath132 is strictly monotone in its @xmath76-th argument with respect to @xmath130 if and only if @xmath133 for all @xmath134 with @xmath135 . from this and from the analogous characterization of @xmath131-monotonicity , it is easy to derive the following lemmata , which will be used in sections [ sect : nrvsq ] and [ sect : incrementality ] .
[ lem : rlinmon ] [ lem : qlinmon ] [ lem : qrlinmon ] for @xmath119 and @xmath136 with @xmath115 , the linear polynomial @xmath137 in @xmath138 $ ] is monotone in all arguments with respect to @xmath122 and well - defined if and only if @xmath139 and @xmath140 for all @xmath74 .
[ lem : qrquadmon ] for @xmath119 and @xmath136 with @xmath115 , the quadratic polynomial @xmath141 in @xmath142 $ ] is monotone with respect to @xmath122 and well - defined if and only if @xmath104 , @xmath143 , @xmath144 , and @xmath145 or @xmath146 . in the remainder of this article we will sometimes use the term `` polynomial interpretations with integer coefficients '' as a synonym for polynomial interpretations over @xmath15 .
likewise , the term `` polynomial interpretations with real ( rational ) coefficients '' refers to polynomial interpretations over @xmath9 ( @xmath14 ) .
[ lucas ] lucas @xcite considers a different definition of polynomial termination over @xmath9 ( @xmath14 ) .
he allows an arbitrary subset @xmath147 ( @xmath148 ) as interpretation domain , provided it is bounded from below and unbounded from above .
the definition of well - definedness is modified accordingly . according to his definition , polynomial termination over @xmath15
trivially implies polynomial interpretations over @xmath9 ( and @xmath14 ) since one can take @xmath149 and @xmath150 , in which case the induced order @xmath151 is the same as the standard order on @xmath15 .
our definitions are based on the understanding that the interpretation domain together with the underlying order determine whether one speaks of polynomial interpretations over the reals , rationals , or integers . as a consequence ,
several of the new results obtained in this paper do not hold in the setting of @xcite .
in this section we show that polynomial termination over @xmath14 implies polynomial termination over @xmath9 .
the proof is based upon the fact that polynomials induce continuous functions , whose behavior at irrational points is completely defined by the values they take at rational points . [ lem : continuity ] let @xmath152 be continuous in @xmath37 . if @xmath153 for all @xmath154 , then @xmath153 for all @xmath155 .
let @xmath156 and let @xmath157 be a sequence of vectors of non - negative rational numbers @xmath158 whose limit is @xmath159 .
such a sequence exists because @xmath160 is dense in @xmath37
. then @xmath161 by continuity of @xmath1 .
thus , @xmath162 is the limit of @xmath163 , which is a sequence of non - negative real numbers by assumption .
hence , @xmath162 is non - negative , too .
[ thm : qvsr ] if a trs is polynomially terminating over @xmath14 , then it is also polynomially terminating over @xmath9 .
let @xmath86 be a trs over the signature @xmath64 that is polynomially terminating over @xmath14 .
so there exists some polynomial interpretation @xmath164 over @xmath14 consisting of a positive rational number @xmath12 and a polynomial @xmath165 $ ] for every @xmath0-ary function symbol @xmath68 such that : 1 . for all @xmath0-ary @xmath68 , @xmath166 for all @xmath167 , 2 . for all @xmath68
, @xmath168 is strictly monotone with respect to @xmath131 in all arguments , 3 .
for every rewrite rule @xmath89 , @xmath169 for all @xmath170 . here
@xmath6 ( @xmath7 ) denotes the polynomial associated with @xmath171 ( @xmath172 ) and the variables @xmath173 are those occurring in @xmath5 .
next we note that all three conditions are quantified polynomial inequalities of the shape `` @xmath174 for all @xmath154 '' for some polynomial @xmath59 with rational coefficients .
this is easy to see for the first and third condition . as to the second condition ,
the function @xmath168 is strictly monotone in its @xmath76-th argument with respect to @xmath131 if and only if @xmath175 for all @xmath176 with @xmath177 , which is equivalent to @xmath178 for all @xmath176 . from lemma [ lem : continuity ] and
the fact that polynomials induce continuous functions we infer that all these polynomial inequalities do not only hold in @xmath113 but also in @xmath112 .
hence , the polynomial interpretation @xmath164 proves termination over @xmath9 .
[ rem : qvsr ] not only does the result established above show that polynomial termination over @xmath14 implies polynomial termination over @xmath9 , but it even reveals that the same interpretation applies .
we conclude this section by showing that for polynomial interpretations over @xmath9 it suffices to consider real _ _ algebraic _ _ numbers as interpretation domain . concerning the use of real algebraic numbers in polynomial interpretations , in ( * ?
* section 6 ) it is shown that it suffices to consider polynomials with real algebraic coefficients as interpretations of function symbols . now the obvious question is whether it is also sufficient to consider only the ( non - negative ) real algebraic numbers @xmath179 instead of the entire set @xmath9 of real numbers as interpretation domain .
we give an affirmative answer to this question by extending the result of @xcite .
a finite trs is polynomially terminating over @xmath9 if and only if it is polynomially terminating over @xmath179 .
let @xmath86 be a trs over the signature @xmath64 that is polynomially terminating over @xmath9 .
there exists a positive real number @xmath12 and a polynomial @xmath114 $ ] for every @xmath0-ary function symbol @xmath68 such that : 1 . for all @xmath0-ary @xmath68 , @xmath180 for all @xmath118 , 2 . for all @xmath68
, @xmath116 is strictly monotone with respect to @xmath130 in all arguments , 3 . for every rewrite rule @xmath89 , @xmath181 for all @xmath182 .
next we treat @xmath12 as a variable and replace all coefficients of the polynomials in @xmath183 by distinct variables @xmath184 .
thus , for each @xmath0-ary function symbol @xmath68 , its interpretation function is a parametric polynomial @xmath185 \subseteq { \mathbb{z}}[{{x_1},\dots,{x_{n}}},{{c_1},\dots,{c_{j}}},\delta]$ ] , where all non - zero coefficients are @xmath186 . as a consequence ,
we claim that all three conditions listed above can be expressed as ( conjunctions of ) quantified polynomial inequalities of the shape @xmath187 for some polynomial @xmath188 $ ] .
this is easy to see for the first condition . for the third condition
it is a direct consequence of the nature of the interpretation functions and the usual closure properties of polynomials . for the second condition we additionally need the fact that @xmath116 is strictly monotone in its @xmath76-th argument with respect to @xmath130 if and only if @xmath189 for all @xmath134
now any of the quantified inequalities can readily be expressed as a formula in the language of ordered fields with coefficients in @xmath24 , where @xmath184 and @xmath12 are the only free variables . by taking the conjunction of all these formulas , existentially quantifying @xmath12 and adding the conjunct @xmath115
, we obtain a formula @xmath190 in the language of ordered fields with free variables @xmath184 and coefficients in @xmath24 ( as @xmath86 and @xmath64 are assumed to be finite ) . by assumption
there are coefficients @xmath191 such that @xmath192 is true in @xmath9 , i.e. , there exists a satisfying assignment for @xmath190 in @xmath9 mapping its free variables @xmath184 to @xmath191 . in order to prove the theorem , we first show that there also exists a satisfying assignment mapping each free variable to a real algebraic number .
we reason as follows . because real closed fields admit quantifier elimination ( ( * ? ? ? * theorem 2.77 ) ) , there exists a quantifier - free formula @xmath193 with free variables @xmath184 and coefficients in @xmath24 that is @xmath9-equivalent to @xmath190 , i.e. , for all @xmath194 , @xmath195 is true in @xmath9 if and only if @xmath196 is true in @xmath9 .
hence , by assumption , @xmath197 is true in @xmath9 .
therefore , the sentence @xmath198 is true in @xmath9 as well .
since both @xmath9 and @xmath179 are real closed fields with @xmath199 and all coefficients in this sentence are from @xmath200 , we may apply the tarski - seidenberg transfer principle ( ( * ? ? ?
* theorem 2.80 ) ) , from which we infer that this sentence is true in @xmath9 if and only if it is true in @xmath179 .
so there exists an assignment for @xmath193 in @xmath179 mapping its free variables @xmath184 to @xmath201 such that @xmath202 is true in @xmath179 , and hence also in @xmath9 as @xmath193 is a boolean combination of atomic formulas in the variables @xmath184 with coefficients in @xmath24 .
but then @xmath203 is true in @xmath9 as well because of the @xmath9-equivalence of @xmath190 and @xmath193 .
another application of the tarski - seidenberg transfer principle reveals that @xmath203 is true in @xmath179 , and therefore the trs @xmath86 is polynomially terminating over @xmath179 ( whose formal definition is the obvious specialization of definition [ def : polyint_r ] ) .
this shows that polynomial termination over @xmath9 implies polynomial termination over @xmath179 .
as the reverse implication can be shown to hold by the same technique , we conclude that polynomial termination over @xmath9 is equivalent to polynomial termination over @xmath179 .
as far as the relationship of polynomial interpretations with real , rational and integer coefficients with regard to termination proving power is concerned , lucas @xcite managed to prove the following two theorems .. ] [ thm : qvsn ] there are trss that are polynomially terminating over @xmath14 but not over @xmath15 .
[ thm : rvsq ] there are trss that are polynomially terminating over @xmath9 but not over @xmath14 or @xmath15 . hence , the extension of the coefficient domain from the integers to the rational numbers entails the possibility to prove some trss polynomially terminating , which could not be proved polynomially terminating otherwise .
moreover , a similar statement holds for the extension of the coefficient domain from the rational numbers to the real numbers . based on these results and the fact that we have the strict inclusions @xmath204 , it is tempting to believe that polynomial interpretations with real coefficients properly subsume polynomial interpretations with rational coefficients , which in turn properly subsume polynomial interpretations with integer coefficients .
indeed , the former proposition holds according to theorem [ thm : qvsr ] .
however , the latter proposition does not hold , as will be shown in this section .
in particular , we present a trs that can be proved terminating by a polynomial interpretation with integer coefficients , but can not be proved terminating by a polynomial interpretation over the reals or rationals . in order to motivate the construction of this particular trs ,
let us first observe that from the viewpoint of number theory there is a fundamental difference between the integers and the real or rational numbers .
more precisely , the integers are an example of a discrete domain , whereas both the real and rational numbers are _ _ dense _ _ and @xmath205 , there exists a real ( rational ) number @xmath45 in between . ] domains . in the context of polynomial interpretations , the consequences of this major distinction are best explained by an example . to this end
, we consider the polynomial function @xmath206 depicted in figure [ fig : nvsrpoly ] and assume that we want to use it as the interpretation of some unary function symbol . now
the point is that this function is permissible in a polynomial interpretation over @xmath15 as it is both non - negative and strictly monotone over the natural numbers .
however , viewing it as a function over a real ( rational ) variable , we observe that non - negativity is violated in the open interval @xmath207 ( and monotonicity requires a properly chosen value for @xmath12 ) .
hence , the polynomial function @xmath206 is not permissible in any polynomial interpretation over @xmath9 ( @xmath14 ) .
( -0.2,0 ) ( 2.4,0 ) node[right ] @xmath30 ; ( 0,-0.1 ) ( 0,3.7 ) node[above ] ; /in 1/1 , 2/2 , 3/3 , 4/4 , 5/5 , 6/6 , 7/7 ( 2pt,0pt ) ( -2pt,0pt ) node[left ] @xmath208 ; /in 0/0 , 1/1 , 2/2 ( 0pt,2pt ) ( 0pt,-2pt ) node[below ] @xmath209 ; ( 0,0 ) parabola bend ( 0.25,-0.125 ) ( 2.1,6.72 ) ; ( 0,0 ) circle ( 0.05 cm ) ; ( 1,0.5 ) circle ( 0.05 cm ) ; ( 2,3 ) circle ( 0.05 cm ) node[below left ] @xmath210 ; thus , the idea is to design a trs that enforces an interpretation of this shape for some unary function symbol , and the tool that can be used to achieve this is polynomial interpolation . to this end , let us consider the following scenario , which is fundamentally based on the assumption that some unary function symbol @xmath211 is interpreted by a quadratic polynomial @xmath212 with ( unknown ) coefficients @xmath213 , @xmath205 and @xmath45 .
then , by polynomial interpolation , these coefficients are uniquely determined by the image of @xmath211 at three pairwise different locations ; in this way the interpolation constraints @xmath214 , @xmath215 and @xmath216 enforce the interpretation @xmath217 .
next we encode these constraints in terms of the trs @xmath86 consisting of the following rewrite rules , where @xmath218 abbreviates @xmath219 , 2 ( ) & ( ) + ^2 ( ) & ( ( ) ) & ( ( ) ) & + ^7 ( ) & ( ^2 ( ) ) & ( ^2 ( ) ) & ^5 ( ) and consider the following two cases : polynomial interpretations over @xmath15 on the one hand and polynomial interpretations over @xmath9 on the other hand . in the context of polynomial interpretations over @xmath15
, we observe that if we equip the function symbols @xmath220 and @xmath221 with the ( natural ) interpretations @xmath222 and @xmath223 , then the trs @xmath86 indeed implements the above interpolation constraints .
is sufficient for this purpose . ] for example , the constraint @xmath224 is expressed by @xmath225 and @xmath226 .
the former encodes @xmath227 , whereas the latter encodes @xmath228 .
moreover , the rule @xmath229 encodes @xmath230 , which is equivalent to @xmath231 in the domain of the natural numbers .
thus , this interpolation constraint can be expressed by a single rewrite rule , whereas the other two constraints require two rules each .
summing up , by virtue of the method of polynomial interpolation , we have reduced the problem of enforcing a specific interpretation for some unary function symbol to the problem of enforcing natural semantics for the symbols @xmath220 and @xmath221 .
next we elaborate on the ramifications of considering the trs @xmath86 in the context of polynomial interpretations over @xmath9 . to this end , let us assume that the symbols @xmath220 and @xmath221 are interpreted by @xmath232 and @xmath233 , so that @xmath220 has some kind of _ successor function _ semantics .
then the trs @xmath86 translates to the following constraints : 2 s_0 - & _ _ ( 0 ) + 2 s_0 - & _ _ ( s_0 ) & _ ( s_0 ) & _ 0 + + 7 s_0 - & _ _ ( 2 s_0 ) & _ ( 2 s_0 ) & _ 5 s_0 + hence , @xmath234 is confined to the closed interval @xmath235 $ ] , whereas @xmath236 is confined to @xmath237 $ ] and @xmath238 to @xmath239 $ ] .
basically , this means that these constraints do not uniquely determine the function @xmath240 .
in other words , the method of polynomial interpolation does not readily apply to the case of polynomial interpretations over @xmath9 .
however , we can make it work .
to this end , we observe that if @xmath241 , then the above system of inequalities actually turns into the following system of equations , which can be viewed as a set of interpolation constraints ( parameterized by @xmath242 ) that uniquely determine @xmath240 : 3 _ ( 0 ) & = 0 & _ ( s_0 ) & = s_0 & _ ( 2 s_0 ) & = 6 s_0 clearly , if @xmath243 , then the symbol @xmath211 is fixed to the interpretation @xmath244 , as was the case in the context of polynomial interpretations over @xmath245 ( note that in the latter case @xmath150 is implicit because of the equivalence @xmath246 ) .
hence , we conclude that once we can manage to design a trs that enforces @xmath241 , we can again leverage the method of polynomial interpolation to enforce a specific interpretation for some unary function symbol .
moreover , we remark that the actual value of @xmath242 is irrelevant for achieving our goal .
that is to say that @xmath242 only serves as a scale factor in the interpolation constraints determining @xmath240 .
clearly , if @xmath247 , then @xmath240 is not fixed to the interpretation @xmath244 ; however , it is still fixed to an interpretation of the same ( desired ) shape , as will become clear in the proof of lemma [ lem : mainlemma1 ] . in the previous subsection
we have presented the basic method that we use in order to show that polynomial interpretations with real or rational coefficients do not properly subsume polynomial interpretations with integer coefficients .
the construction presented there was based on several assumptions , the essential ones of which are : 1 .
[ ass1 ] the symbol @xmath220 had to be interpreted by a linear polynomial of the shape @xmath248 .
[ ass2 ] the condition @xmath241 was required to hold .
[ ass3 ] the function symbol @xmath211 had to be interpreted by a quadratic polynomial .
now the point is that one can get rid of all these assumptions by adding suitable rewrite rules to the trs @xmath86 .
the resulting trs will be referred to as @xmath20 , and it consists of the rewrite rules given in table [ tab : trs_r1 ] . the rewrite rules and serve the purpose of ensuring the first of the above items .
informally , constrains the interpretation of the symbol @xmath220 to a linear polynomial by simple reasoning about the degrees of the left- and right - hand side polynomials , and does the same thing with respect to @xmath249 .
because both interpretations are linear , compatibility with can only be achieved if the leading coefficient of the interpretation of @xmath220 is one .
concerning item above , we remark that the tricky part is to enforce the upper bound of two on the degree of the polynomial @xmath240 that interprets the symbol @xmath211 . to this end , we make the following observation
. if @xmath240 is at most quadratic , then the function @xmath250 is at most linear ; i.e. , there is a linear function @xmath251 such that @xmath252 , or equivalently , @xmath253 , for all values of @xmath30 .
this can be encoded in terms of rule as soon as the interpretation of @xmath254 corresponds to addition of two numbers . and
this is exactly the purpose of rules , and .
more precisely , by linearity of the interpretation of @xmath249 , we infer from that the interpretation of @xmath254 must have the linear shape @xmath255 .
furthermore , compatibility with and implies @xmath256 due to item above .
hence , the interpretation of @xmath254 is @xmath257 , and it really models addition of two numbers ( modulo adding a constant ) .
next we comment on how to enforce the second of the above assumptions .
to this end , we remark that the hard part is to enforce the condition @xmath258 .
the idea is as follows .
first , we consider rule , observing that if @xmath211 is interpreted by a quadratic polynomial @xmath240 and @xmath220 by the linear polynomial @xmath248 , then ( the interpretation of ) its right - hand side will eventually become larger than its left - hand side with growing @xmath242 , thus violating compatibility . in this way
, @xmath242 is bounded from above , and the faster the growth of @xmath240 , the lower the bound . the problem with this statement ,
however , is that it is only true if @xmath240 is fixed ( which is a priori not the case ) ; otherwise , for any given value of @xmath242 , one can always find a quadratic polynomial @xmath240 such that compatibility with is satisfied .
the parabolic curve associated with @xmath240 only has to be flat enough .
so in order to prevent this , we have to somehow control the growth of @xmath240 .
now that is where rule comes into play , which basically expresses that if one increases the argument of @xmath240 by a certain amount ( i.e. , @xmath259 ) , then the value of the function is guaranteed to increase by a certain minimum amount , too .
thus , this rule establishes a lower bound on the growth of @xmath240 . and it turns out that if @xmath240 has just the right amount of growth , then we can readily establish the desired upper bound @xmath12 for @xmath242 . finally , having presented all the relevant details of our construction , it remains to formally prove our main claim that the trs @xmath20 is polynomially terminating over @xmath15 but not over @xmath9 or @xmath14 .
the trs @xmath20 is polynomially terminating over @xmath15 .
we consider the following interpretation : @xmath260 note that the polynomial @xmath244 is a permissible interpretation function as it is both non - negative and strictly monotone over the natural numbers by lemma [ lem : nquadmon ] ( cf .
figure [ fig : nvsrpoly ] ) .
the rewrite rules of @xmath20 are compatible with this interpretation because the resulting inequalities 2 1 & > _ 0 & 32x^2 + 60x + 28 & > _ 32x^2 - 16x + 20 + 2 & > _ 1 & 4x + 8 & > _ 4x + 6 + 7 & > _ 6 & 4x+4 & > _ 2x + 1 & > _ 0 & x+1 & > _ x + 6 & > _ 5 & x+1 & > _ x + 2x^2 + 7x + 6 & > _ 2x^2 + 7x + 4 & 2x^2 + 3x + 4 & > _ 2x^2 + 3x + 1 are clearly satisfied for all natural numbers @xmath30 .
[ lem : mainlemma1 ] the trs @xmath20 is not polynomially terminating over @xmath9 .
let us assume that @xmath20 is polynomially terminating over @xmath9 and derive a contradiction .
compatibility with rule implies @xmath261 as a consequence , @xmath262 , and because @xmath263 and @xmath264 must be strictly monotone , we conclude @xmath265 .
the same reasoning applied to rule yields @xmath266 . hence
, the symbols @xmath220 and @xmath249 must be interpreted by linear polynomials .
so @xmath267 and @xmath268 with @xmath269 and , due to lemma [ lem : rlinmon ] , @xmath270 and @xmath271
. then the compatibility constraint imposed by rule gives rise to the inequality @xmath272 which must hold for all non - negative real numbers @xmath30 .
this implies the following condition on the respective leading coefficients : @xmath273 .
because of @xmath270 and @xmath271 , this can only hold if @xmath274 . hence , @xmath275 .
this result simplifies to @xmath276 , which implies @xmath277 . from this
, we conclude that @xmath278 and @xmath279 .
now suppose that the function symbol @xmath211 were also interpreted by a linear polynomial @xmath240 .
then we could apply the same reasoning to rule because it is structurally equivalent to , thus inferring @xmath280 . however , this would contradict @xmath279 ; therefore , @xmath240 can not be linear .
next we turn our attention to the rewrite rules , and . because @xmath264 is linear , compatibility with constrains the function @xmath281 to be at most linear .
this can only be the case if @xmath282 contains no terms of degree two or higher .
in other words , @xmath283 , where @xmath284 , @xmath285 and @xmath286 ( cf .
lemma [ lem : rlinmon ] ) . because of @xmath275 , compatibility with implies @xmath287 , and compatibility with implies @xmath288
; thus , @xmath289 . using the obtained information in the compatibility constraint associated with rule
, we get @xmath290 this implies that @xmath291 , which simplifies to @xmath292 because @xmath293 . consequently
, @xmath240 must be a quadratic polynomial . without loss of generality ,
let @xmath294 , subject to the constraints : @xmath295 and @xmath296 because of non - negativity ( for all @xmath297 ) , and @xmath298 because @xmath299 due to strict monotonicity of @xmath240 .
next we consider the compatibility constraint associated with rule , from which we deduce an important auxiliary result .
after unraveling the definitions of @xmath130 and the interpretation functions , this constraint simplifies to @xmath300 which implies the following condition on the respective leading coefficients : @xmath301 ; from this and @xmath279 , we conclude @xmath302 and note that @xmath303 . hence , @xmath304 expresses the change of the slopes of the tangents to @xmath240 at the points @xmath305 and @xmath306 , and thus actually sets a lower bound on the growth of @xmath240 .
now let us consider the combined compatibility constraint imposed by the rules and , namely @xmath307 , which implies @xmath308 by definition of @xmath130 .
thus , we conclude @xmath309 .
in fact , we even have @xmath241 , which can be derived from the compatibility constraint of rule using the conditions @xmath309 , @xmath298 and @xmath310 , the combination of the former two conditions : @xmath311 hence , @xmath312 , or equivalently , @xmath313 . but because of and @xmath309 , this inequality can only be satisfied if : @xmath314 this result has immediate consequences concerning the interpretation of the constant @xmath221 . to this end
, we consider the compatibility constraint of rule , which simplifies to @xmath315 .
because of and the fact that @xmath316 and @xmath317 must be non - negative , we conclude @xmath318
. moreover , condition is the key to the proof of this lemma . to this end
, we consider the compatibility constraints associated with the five rewrite rules : 2 s_0 & > _ _ 0,s_0 _ ( 0 ) + 2 s_0 & > _ _ 0,s_0 _ ( s_0 ) & _ ( s_0 ) & > _ _ 0,s_0 0 + 7 s_0 & > _ _ 0,s_0 _ ( 2 s_0 ) & _ ( 2 s_0 ) & > _ _ 0,s_0 5 s_0 by definition of @xmath319 , these inequalities give rise to the following system of equations : 3 _ ( 0 ) & = 0 & _ ( s_0 ) & = s_0 & _ ( 2 s_0 ) & = 6 s_0 after unraveling the definition of @xmath240 and substituting @xmath320 , we get a system of linear equations in the unknowns @xmath321 , @xmath205 and @xmath45 3 c & = 0 & z + b & = 1 & 4z + 2b & = 6 which has the unique solution @xmath322 , @xmath323 and @xmath324 .
hence , @xmath240 must have the shape @xmath325 in every compatible polynomial interpretation over @xmath9 .
however , this function is not a permissible interpretation for the function symbol @xmath211 because it is not non - negative for all @xmath297 . in particular
, it is negative in the open interval @xmath326 ; e.g. @xmath327 .
hence , @xmath20 is not compatible with any polynomial interpretation over @xmath9 . in this proof
the interpretation of @xmath211 is fixed to @xmath328 , which violates well - definedness in @xmath112 . however ,
this function is obviously well - defined in @xmath110 for a properly chosen negative real number @xmath329 .
so what happens if we take this @xmath110 instead of @xmath112 as the carrier of a polynomial interpretation ? to this end , we observe that @xmath330 and @xmath331 .
now let us consider some negative real number @xmath332 .
we have @xmath333 and thus @xmath334 , which means that @xmath240 violates monotonicity with respect to the order @xmath335 . the previous lemma , together with theorem [ thm : qvsr ] , yields the following corollary .
the trs @xmath20 is not polynomially terminating over @xmath14 . finally , combining the material presented in this section , we establish the following theorem , the main result of this section .
[ thm : nvsr ] there are trss that can be proved polynomially terminating over @xmath15 , but can not be proved polynomially terminating over @xmath9 or @xmath14 .
we conclude this section with a remark on the actual choice of the polynomial serving as the interpretation of the function symbol @xmath211 .
as explained at the beginning of this section , the trs @xmath20 was designed to enforce an interpretation for @xmath211 , which is permissible in a polynomial interpretation over @xmath15 but not over @xmath9 ( @xmath14 ) .
the interpretation of our choice was the polynomial @xmath244 .
however , we could have chosen any other polynomial as long as it is well - defined and strictly monotone over @xmath15 but not over @xmath9 ( @xmath14 ) .
the methods introduced in this section are general enough to handle any such polynomial .
so the actual choice is not that important .
this section is devoted to showing that polynomial termination over @xmath15 and @xmath9 does not imply polynomial termination over @xmath14 .
the proof is constructive , so we give a concrete trs having the desired properties . in order to motivate the construction underlying this particular system ,
let us consider the following quantified polynomial inequality @xmath336 where @xmath337 $ ] is a polynomial with integer coefficients , all of whose roots are irrational and which is positive for some non - negative integer value of @xmath213 . to be concrete ,
let us take @xmath338 and try to satisfy in @xmath15 , @xmath113 and @xmath112 , respectively .
first , we observe that @xmath339 is a satisfying assignment in @xmath112 .
moreover , is also satisfiable in @xmath15 by assigning @xmath340 , for example , and observing that the polynomial @xmath341 is non - negative for all @xmath342 .
however , can not be satisfied in @xmath113 as non - negativity of @xmath341 does not hold for all @xmath343 and @xmath59 has no rational roots . to sum up ,
is satisfiable in @xmath15 and @xmath112 but not in @xmath113 .
thus , the basic idea now is to design a trs containing some rewrite rule whose compatibility constraint reduces to a polynomial inequality similar in nature to . to this end , we rewrite the inequality @xmath344 to @xmath345 because now both the left- and right - hand side can be viewed as a composition of several functions , each of which is strictly monotone and well - defined . in particular , we identify the following constituents : @xmath346 , @xmath347 , @xmath348 and @xmath349 .
thus , the above inequality can be written in the form @xmath350 which can easily be modeled as a rewrite rule .
( note that @xmath351 is not strictly necessary as @xmath352 , but it gives rise to a shorter encoding . ) and then we also need rewrite rules that enforce the desired interpretations for the function symbols @xmath254 , @xmath353 , @xmath354 and @xmath355 . for this purpose ,
we leverage the techniques presented in the previous section , in particular the method of polynomial interpolation .
the resulting trs will be referred to as @xmath21 , and it consists of the rewrite rules given in table [ tab : trs_r2 ] .
.the trs @xmath21 . [ cols="^,^ " , ] as in section [ sect : nrvsq ] , each block serves a specific purpose .
the one made up of enforces the desired semantics for the symbol @xmath353 , that is , a linear function @xmath356 that doubles its input , while the block enforces a linear function @xmath357 for the symbol @xmath358 whose slope @xmath359 is proportional to the square of the interpretation of the constant @xmath360 .
finally , encodes the main idea of the construction , as mentioned above .
[ lem : trs_r_5_sn ] the trs @xmath361 is incrementally polynomially terminating over @xmath15 and @xmath9 . for incremental polynomial termination over @xmath15 , we start with the interpretation @xmath362 all interpretation functions are well - defined over @xmath15 and strictly monotone ( i.e. , monotone with respect to @xmath95 ) as well as weakly monotone ( i.e. , monotone with respect to @xmath99 ) . moreover , it is easy to verify that this interpretation is weakly compatible with @xmath361 . in particular ,
the rule gives rise to the constraint @xmath363 which holds for all @xmath342 .
after removing the rules from @xmath361 for which ( strict ) compatibility holds , we are left with the rules , , , and , all of which can be handled ( that is , removed at once ) by the following linear interpretation : @xmath364 for incremental polynomial termination over @xmath9 , we consider the interpretation @xmath365 which is both weakly and strictly monotone according to lemmata [ lem : qrquadmon ] and [ lem : qrquadwmon ] .
so all interpretation functions are well - defined over @xmath112 and monotone with respect to @xmath130 and @xmath366 .
moreover , one easily verifies that this interpretation is weakly compatible with @xmath361 .
in particular , the constraint @xmath367 associated with trivially holds . after removing the rules from @xmath361 for which ( strict ) compatibility holds ( i.e. , for which the corresponding constraint remains true after strengthening @xmath366 to @xmath130 ) ,
we are left with , , , and , all of which can be removed at once by the following linear interpretation : @xmath368 [ lem : trs_r_5 ] the trs @xmath361 is not incrementally polynomially terminating over @xmath14 .
let @xmath369 be a strictly monotone polynomial interpretation over @xmath14 that is weakly compatible with @xmath361 . according to lemma [ lem : trs_s ] ,
the symbols @xmath221 , @xmath220 , @xmath211 , @xmath249 , @xmath254 and @xmath355 are interpreted as follows : @xmath370 where @xmath371 and @xmath372 .
as the interpretation of @xmath355 is linear , weak compatibility with the rule implies that the interpretation of @xmath353 is at most linear as well , i.e. , @xmath373 with @xmath374 and @xmath375 .
we also have @xmath376 due to weak compatibility with and @xmath377 due to weak compatibility with ; hence , @xmath378 . similarly , by linearity of @xmath379 and weak compatibility with
, the interpretation of @xmath358 must have the shape @xmath380 .
then weak compatibility with yields @xmath381 ; hence , @xmath382 , @xmath383 .
next we note that weak compatibility with and implies that @xmath384 , which evaluates to @xmath385 . from this
we infer that @xmath386 as @xmath104 and @xmath387 due to weak compatibility with ; i.e. , @xmath388 .
finally , we consider the weak compatibility constraint associated with , which simplifies to @xmath389 however , this inequality is unsatisfiable as the polynomial @xmath390 is negative for some @xmath343 and @xmath391 can not be zero because both @xmath213 and @xmath392 must be rational numbers .
so we conclude that there is no strictly monotone polynomial interpretation over @xmath14 that is weakly compatible with the trs @xmath361 .
this implies that @xmath361 is not incrementally polynomially terminating over @xmath14 .
together , lemma [ lem : trs_r_5_sn ] and lemma [ lem : trs_r_5 ] yield the main result of this subsection .
[ cor : nrvsq_inc ] there are trss that are incrementally polynomially terminating over @xmath15 and @xmath9 but not over @xmath14 . in this subsection , we show that there are trss that are incrementally polynomially terminating over @xmath15 but not over @xmath9 . for this purpose ,
we extend the trs @xmath393 of table [ tab : trs_s ] by the single rewrite rule @xmath394 and call the resulting system @xmath395 .
[ lem : trs_r_6_sn ] the trs @xmath395 is incrementally polynomially terminating over @xmath15 .
first , we consider the interpretation @xmath396 which is both weakly and strictly monotone as well as weakly compatible with @xmath395 .
in particular , the constraint @xmath397 associated with @xmath398 holds for all @xmath342 .
removing the rules from @xmath395 for which ( strict ) compatibility holds leaves us with the rules , and @xmath398 , which are easily handled , e.g. by the linear interpretation @xmath399 [ lem : trs_r_6 ] the trs @xmath395 is not incrementally polynomially terminating over @xmath9 or @xmath14 .
let @xmath119 , and let @xmath369 be a polynomial interpretation over @xmath129 that is weakly compatible with @xmath395 , and in which the interpretation of the function symbol @xmath211 has the shape @xmath400 with @xmath104 .
then the weak compatibility constraint @xmath401 associated with @xmath398 does not hold for all @xmath402 because the polynomial @xmath403 is negative in the open interval @xmath404 . as the above assumption on the interpretation of @xmath211 follows from lemma [ lem : trs_s ]
if @xmath369 is strictly monotone , we conclude that there is no strictly monotone polynomial interpretation over @xmath9 or @xmath14 that is weakly compatible with the trs @xmath395 .
this implies that @xmath395 is not incrementally polynomially terminating over @xmath9 or @xmath14 .
together , lemma [ lem : trs_r_6_sn ] and lemma [ lem : trs_r_6 ] yield the main result of this subsection . [ cor : nvsr_inc ] there are trss that are incrementally polynomially terminating over @xmath15 but not over @xmath9 or @xmath14 .
the results presented in this section can be summarized by stating that the relationships expressed in figure [ fig : summary ] remain true for incremental polynomial termination , after replacing @xmath20 by @xmath395 and @xmath21 by @xmath361 .
in this article , we investigated the relationship of polynomial interpretations with real , rational and integer coefficients with respect to termination proving power .
in particular , we presented three new results , the first of which shows that polynomial interpretations over the reals subsume polynomial interpretations over the rationals , the second of which shows that polynomial interpretations over the reals or rationals do not properly subsume polynomial interpretations over the integers , a result that comes somewhat unexpected , and the third of which shows that there are trss that can be proved terminating by polynomial interpretations over the naturals or the reals but not over the rationals .
these results were extended to incremental termination proofs . in @xcite it
is shown how to adapt the results to the dependency pair framework @xcite .
we conclude this article by reviewing our results in the context of automated termination analysis , where linear polynomial interpretations , i.e. , polynomial interpretations with all interpretation functions being linear , play an important role .
this naturally raises the question as to what extent the restriction to linear polynomial interpretations influences the hierarchy depicted in figure [ fig : summary ] , and in what follows we shall see that it changes considerably .
more precisely , the areas inhabited by the trss @xmath20 and @xmath21 become empty , such that polynomial termination by a linear polynomial interpretation over @xmath15 implies polynomial termination by a linear polynomial interpretation over @xmath14 , which in turn implies polynomial termination by a linear polynomial interpretation over @xmath9 .
the latter follows directly from theorem [ thm : qvsr ] and remark [ rem : qvsr ] , whereas the former is shown below .
polynomial termination by a linear polynomial interpretation over @xmath15 implies polynomial termination by a linear polynomial interpretation over @xmath14 .
let @xmath86 be a trs that is compatible with a linear polynomial interpretation @xmath164 over @xmath15 , where every @xmath0-ary function symbol @xmath211 is associated with a linear polynomial @xmath405 .
we show that the same interpretation also establishes polynomial termination over @xmath14 with the value of @xmath12 set to one . to this end
, we note that in order to guarantee strict monotonicity and well - definedness over @xmath15 , the coefficients of the respective interpretation functions have to satisfy the following conditions : @xmath406 and @xmath407 for all @xmath408 .
hence , by lemma [ lem : qlinmon ] , we also have well - definedness over @xmath113 and strict monotonicity with respect to the order @xmath409 .
( strict monotonicity also follows from ( * ? ? ?
* theorem 2 ) . ) moreover , as @xmath86 is compatible with @xmath164 , each rewrite rule @xmath89 satisfies @xmath410 where @xmath6 ( @xmath7 ) denotes the polynomial associated with @xmath171 ( @xmath172 ) and the variables @xmath173 are those occurring in @xmath5 . since linear functions are closed under composition , the polynomial @xmath411 is a linear polynomial @xmath412 , such that holds if and only if @xmath413 and @xmath414 for all @xmath415 .
however , then we also have @xmath416 which shows that @xmath86 is compatible with the linear polynomial interpretation @xmath417 over @xmath14 .
hence , linear polynomial interpretations over @xmath9 subsume linear polynomial interpretations over @xmath14 , which in turn subsume linear polynomial interpretations over @xmath15 , and these subsumptions are proper due to the results of @xcite , which were obtained using linear polynomial interpretations .
we thank harald zankl for finding the incremental polynomial interpretation given in the proof of lemma [ lem : r1_ipt_q ] .
the comments by the reviewers improved the presentation and helped to clarify the contributions of salvador lucas @xcite .
d. hofbauer
. termination proofs by context - dependent interpretations . in _ proc .
12th international conference on rewriting techniques and applications ( rta 2001 ) _ , volume 2051 of _ lecture notes in computer science _ , pages 108121 , 2001 .
http://dx.doi.org/10.1007/3-540-45127-7_10[doi : ] .
s. lucas . on the relative power of polynomials with real , rational , and integer coefficients in proofs of termination of rewriting .
, 17(1):4973 , 2006 .
http://dx.doi.org/10.1007/s00200-005-0189-5[doi : ] .
s. lucas .
practical use of polynomials over the reals in proofs of termination . in _ proc .
9th international conference on principles and practice of declarative programming ( ppdp 2007 ) _ , pages 3950 .
association of the computing machinery , 2007 .
http://dx.doi.org/10.1145/1273920.1273927[doi : ] .
f. neurauter and a. middeldorp .
polynomial interpretations over the reals do not subsume polynomial interpretations over the integers . in _ proc .
21st international conference on rewriting techniques and applications ( rta 2010 ) _ , volume 6 of _ leibniz international proceedings in informatics _ , pages 243258 , 2010 .
http://dx.doi.org/10.4230/lipics.rta.2010.243[doi : ] . | polynomial interpretations are a useful technique for proving termination of term rewrite systems .
they come in various flavors : polynomial interpretations with real , rational and integer coefficients . as to their relationship with respect to termination proving power
, lucas managed to prove in 2006 that there are rewrite systems that can be shown polynomially terminating by polynomial interpretations with real ( algebraic ) coefficients , but can not be shown polynomially terminating using polynomials with rational coefficients only .
he also proved the corresponding statement regarding the use of rational coefficients versus integer coefficients . in this article
we extend these results , thereby giving the full picture of the relationship between the aforementioned variants of polynomial interpretations .
in particular , we show that polynomial interpretations with real or rational coefficients do not subsume polynomial interpretations with integer coefficients .
our results hold also for incremental termination proofs with polynomial interpretations . |
a resonance is fundamentally a non - perturbative object and is thus not amenable to the methods of standard perturbation theory . in order to describe physics at the @xmath1 resonance one
is forced , therefore , to employ some sort of non - perturbative procedure .
such a procedure is dyson summation that sums strings of one - particle irreducible ( 1pi ) self - energy diagrams as a geometric series to all orders in the coupling constant , @xmath3 and effectively replaces the tree - level @xmath1 propagator by a dressed propagator , @xmath4 where @xmath5 is the one - loop @xmath1 self - energy .
the problem here is that electroweak physics is described by a gauge theory .
results of calculations of physical processes must be exactly gauge - invariant but this comes about through delicate cancellations between many different feynman diagrams each of which is separately gauge - dependent .
the cancellation happens at each order in @xmath3 when all diagrams of a given order are combined .
the @xmath1 self - energy , @xmath5 , is gauge - dependent at @xmath6 and hence the rhs of eq.([eq : dysonsum ] ) is gauge - dependent at all orders in @xmath3 .
if the dressed propagator is used in a finite - order calculation the result will be gauge - dependent at some order because the will be no diagrams available to cancel the gauge - dependence beyond the order being calculated .
this gauge - dependence should be viewed as an indicator that the approximation scheme being used is inconsistent or does not represent a physical observable . in constructing amplitudes that represent physical observables
, one must take care to respect the requirements that are laid down by analytic @xmath0-matrix theory @xcite .
these conditions are derived from general considerations such as energy conservation and causality and it will be found that appealing to them leads to a procedure for generating gauge - invariant amplitudes with no flexibility in the final result .
the procedure will also make it possible to give an answer to the _ problem of the unstable particle _ that was put forward by peierls in the early fifties @xcite . much of what appears here can be found in ref.s@xcite .
we will begin by reviewing what is known about @xmath0-matrix elements for processes containing unstable particles .
an unstable particle is associated with a pole , @xmath7 , lying on the second riemann sheet below the real @xmath8 axis .
the scattering amplitude , @xmath9 for a process which contains an intermediate unstable particle can then be written in the form , @xmath10 where @xmath11 and @xmath7 are complex constants and @xmath12 is regular at @xmath13 . the first term on the rhs of eq.([eq : analyticamp ] )
will be called the resonant term and the second is the non - resonant background term . it is known that @xmath7 is process - independent in the sense that any process that contains a given unstable particle as an intermediate state will have its pole at the same position .
it can also be shown from fredholm theory that the residue factorizes as , @xmath14 , into pieces that depend separately on the initial- and final - state .
one can prove , by very simple arguments , that @xmath7 , @xmath11 and @xmath12 are separately and exactly gauge - invariant @xcite .
> from analytic @xmath0-matrix theory it is known that production thresholds are associated with a branch cut .
branch points for stable particles lie on the real @xmath8-axis and those for unstable particles , such as @xmath15 bosons , lie significantly below it .
provided there are no nearby thresholds the amplitude , @xmath9 , can be adequately described in the resonance region by a laurent expansion about the pole , @xmath7 .
it should be emphasized that @xmath9 can always be written in the form ( [ eq : analyticamp ] ) even when thresholds are present . in that case @xmath12
will be an analytic function containing a branch point and the rhs of ( [ eq : analyticamp ] ) continues to be an exact representation of @xmath9 .
laurent expansion refers to how the resonant term is identified .
the aim is only to separate @xmath9 into gauge - invariant resonant and background pieces and there is no necessity to perform a laurent expansion beyond its leading term although this can provide a useful way for parameterizing electroweak data @xcite .
let us consider the production process for a massless fermion pair in @xmath16 .
away from the @xmath1 resonance it is known how to calculate the amplitude to arbitrary accuracy in a gauge - invariant manner . near resonance
we must perform a dyson summation and then separate the amplitude into its resonant and background pieces .
doing so allows expressions for the gauge - invariant quantities , @xmath7 , @xmath11 and @xmath12 to be identified in terms of the 1pi functions that occur in perturbation theory .
the scattering amplitude for @xmath16 is @xmath17 in which @xmath18f_{zz}^{\frac{1}{2}}(s),\\ r_{zf}(s)&=&f_{zz}^{\frac{1}{2}}(s)\left[v_{zf}(s ) + \frac{\pi_{z\gamma}(s)}{s-\pi_{\gamma\gamma}(s ) } v_{\gamma f}(s)\right ] .
\label{eq : residuefactor}\end{aligned}\ ] ] the pole position @xmath7 is a solution of the equation @xmath19 and @xmath20 is defined through the relation @xmath21 it should be emphasized that eq.([eq : exactamp ] ) is exact and valid anywhere on the complex @xmath8-plane .
the effect of @xmath1-@xmath22 mixing has been included .
the quantity @xmath23 and @xmath24 are the photon self - energy and the @xmath25-@xmath22 mixing respectively .
@xmath26 , @xmath27 are the initial- and final - state @xmath1 vertices , into which the external wavefunctions have been absorbed , and @xmath28 , @xmath29 are the corresponding vertices for the photon . here
@xmath30 denotes 1pi corrections to the matrix element that include things like as box diagrams . the first term on the rhs of eq.([eq : exactamp ] )
is the resonant part of @xmath31 and the three terms on the second line taken together are form the non - resonant background .
calculations are most conveniently performed in terms of the real renormalized parameters of the theory such as the renormalized mass , @xmath32 .
eq.([eq : poleeqn ] ) can be solved iteratively in terms of @xmath32 to give @xmath33 the rhs of eq.([eq : polesoln ] ) may be substituted for @xmath7 where it appears in eq.([eq : exactamp])([eq : residuefactor ] ) .
taylor series expansion can then be used to obtain perturbative expressions for @xmath7 , @xmath11 and @xmath12 in terms of greens functions with real arguments up to any desired order . at any given order
these perturbative expressions will be exactly gauge - invariant as will scattering amplitudes constructed from them .
a couple of points should be noted here . the appearance of greens functions with complex arguments in eq.([eq : exactamp ] ) is a natural consequence of the analyticity of the @xmath0-matrix .
the @xmath0-matrix itself is never evaluated with complex @xmath8 .
the arguments of @xmath31 on the lhs of eq.([eq : exactamp ] ) is real as is the ` @xmath8 ' in the denominator of the first term on the rhs . in the procedure described above
, one starts by extracting the resonant term in a scattering amplitude by laurent expansion about the exact pole position @xmath7 and then specializes to lower orders by further expanding about the renormalized mass .
other authors @xcite have attempted to apply the techniques described above by first expanded about the renormalized mass and then added a finite width in the denominator of the resonant part by hand . that procedure can not be justified and leads to problems when one treats processes like @xmath34 .
it gives rise to spurious _ threshold singularities _ or complex scattering angles due the production threshold s being incorrectly located on the real axis . in section iv the process @xmath35 will be treated and no threshold singularities or complex scattering angles will arise .
we have thus succeeded in our goal of producing exactly gauge - invariant scattering amplitudes to arbitrary order .
one might ask whether what has been done is just a mathematical trick , in which case the gauge - invariance is accidental , or does it have some physical interpretation . in this section
it will be shown that the latter is true .
recall that the coordinate space dressed propagator for a scalar particle has an integral representation @xmath36 the integrand has a pole at @xmath37 where @xmath7 is a solution of the equation @xmath38 and as in the previous section we define @xmath39 via the relation @xmath40 .
the dressed propagator can then be written as @xmath41 \label{eq : splitprop}\ ] ] that separates resonant and non - resonant pieces . performing the @xmath42 integration resonant
gives @xmath43 where @xmath44 .
the non - resonant term contributes only for @xmath45 and so represents a contact interaction .
the resonant part spits into two terms that contribute when @xmath46 or @xmath47 and therefore connects points @xmath48 and @xmath49 that are separated by a finite distance in space - time .
the problem of the unstable particle @xcite is may be roughly stated as follows : @xmath0-matrix theory deals with asymptotic in - states and out - states that propagate from and to infinity .
unstable particles can not exist as asymptotic states because they decay a finite distance from the interaction region .
indeed it is known @xcite that the @xmath0-matrix is unitary on the hilbert space spanned by stable particle states and hence there is not even any room to accommodate unstable particles as external states .
how can one use the @xmath0-matrix to calculate , say , the production cross - section for an unstable particle when it can not exist as an asymptotic state ? in the first part of this section it was shown that the resonant part of the dressed propagator connected points with a finite space - time separation .
when a similar analysis is applied to a physical matrix element , such as eq.([eq : exactamp ] ) , one concludes that the resonant part describes a process in which there is a finite space - time separation between the initial - state vertex , @xmath50 , and the final - state vertex , @xmath51 . in other words ,
the resonant term describes the finite propagation of a physical @xmath1 boson .
the non - resonant background represents prompt production of the final state
. as these two possibilities are , in principle , physically distinguishable , they must be separately gauge - invariant .
we can thus use finite propagation as a tag for identifying unstable particles without requiring that they appear in the final state .
this is , after all , the way @xmath52-quarks are identified in vertex detectors .
a production cross - section for an unstable particle is obtained by extracting the resonant part of the matrix element for a process containing that particle in an intermediate state and summing over all possible decay modes .
in this section we will calculate the cross - section for @xmath35 .
this is of both theoretical and practical importance . on the theoretical side it represents an example of a calculation of the production cross - section for unstable particles . on the practical side , at high energies @xmath35 will be a dominant source of fermion pairs @xmath54 and @xmath55 due to its double resonant enhancement and hence @xmath2 is an excellent approximation to the cross - section for 4-fermion pair production .
if the experimental situation warrants it , background terms can also be included without difficulty . in the case of @xmath16 , dealt with in section ii , the invariant mass squared of the @xmath1 , @xmath8 , is fixed by the momenta of the incoming @xmath56 . for the process @xmath35
the invariant mass of the produced @xmath1 s is not constant and must be somehow included in phase space integrations .
it is not immediately clear how to do this and without further guidance from @xmath0-matrix theory there would seem to be considerable flexibility in how to proceed .
a new ingredient is required and that is to realize that an expression for an @xmath0-matrix element can always be divided into a part that is a lorentz - invariant function of the kinematic invariants of the problem and lorentz - covariant objects , such as @xmath57 etc .
the latter are known as _ standard covariants _ @xcite .
it is the lorentz invariant part that satisfies the requirements of analytic @xmath0-matrix theory and from which the resonant and non - resonant background parts are extracted while the lorentz covariant part is untouched . to calculate @xmath2
, we begin by constructing the cross - section for @xmath58 , and will eventually sum over all fermion species .
the part of the full matrix element that can give rise to doubly resonant contributions can be written as @xmath59 m_i(t , u , p_1 ^ 2,p_2 ^ 2)\nonumber\\ & & \ \ \ \ \ \ \
\times\frac{1}{p_1 ^ 2-m_z^2-\pi_{zz}(p_1 ^ 2 ) } [ \bar u_{f_1}\gamma^\mu(v_{zf_l}(p_1 ^ 2)\gamma_l + v_{zf_r}(p_1 ^
2)\gamma_r ) v_{\bar f_1 } ] \label{eq : fullzz}\\ & & \ \ \ \ \ \ \ \times\frac{1}{p^2_2-m_z^2-\pi_{zz}(p_2 ^ 2 ) } [ \bar u_{f_2}\gamma^\nu(v_{zf_l}(p_2 ^ 2)\gamma_l + v_{zf_r}(p_2 ^ 2)\gamma_r ) v_{\bar f_2 } ] \nonumber\end{aligned}\ ] ] where @xmath60 are lorentz covariant tensors that span the tensor structure of the matrix element and @xmath61 , @xmath62 are the usual helicity projection operators
. the squared invariant masses of the @xmath63 and @xmath64 pairs are @xmath65 and @xmath66 .
the @xmath67 , @xmath68 and @xmath69 are lorentz scalars that are analytic functions of the independent kinematic lorentz invariants of the problem . to extract the piece of the matrix element that corresponds to finite propagation of both @xmath1 s we extract the leading term in a laurent expansion in @xmath65 and @xmath66 of the analytic lorentz - invariant part of eq.([eq : fullzz ] ) leaving the lorentz - covariant part untouched .
this is the doubly - resonant term and is given by @xmath59 m_i(t , u
, s_p , s_p)\nonumber\\ & & \ \ \ \ \ \ \
\times\frac{f_{zz}(s_p)}{p_1 ^ 2-s_p } [ \bar
u_{f_1}\gamma^\mu(v_{zf_l}(s_p)\gamma_l + v_{zf_r}(s_p)\gamma_r ) v_{\bar f_1 } ] \label{eq : reszz}\\ & & \ \ \ \ \ \ \ \times\frac{f_{zz}(s_p)}{p_2 ^ 2-s_p } [ \bar
u_{f_2}\gamma^\nu(v_{zf_l}(s_p)\gamma_l + v_{zf_r}(s_p)\gamma_r ) v_{\bar f_2 } ] \nonumber\end{aligned}\ ] ] where @xmath70 defined by a relation like ( [ eq : fzzdef ] ) .
it should be emphasized that eq.([eq : reszz ] ) is the exact form of the doubly - resonant matrix element to all orders in perturbation theory that we will now specialize to leading order .
it is free of threshold singularities noted that were found by other authors @xcite . in lowest order eq.([eq : reszz ] ) becomes , up to overall multiplicative factors , @xmath71m_i\nonumber\\ & & \ \ \ \
\times\frac{1}{p_1 ^ 2-s_p } [ \bar u_{f_1}\gamma^\mu(v_{zf_l}\gamma_l+v_{zf_r}\gamma_r)v_{\bar f_1 } ] \label{eq : lowestzz}\\ & & \ \ \ \ \times\frac{1}{p_2
^ 2-s_p } [ \bar u_{f_2}\gamma^\nu(v_{zf_l}\gamma_l+v_{zf_r}\gamma_r)v_{\bar f_2 } ] . \nonumber\end{aligned}\ ] ] where @xmath72 , @xmath73 ; @xmath74 , @xmath75 and the final state vertex corrections take the form @xmath76 and @xmath77 , the left- and right - handed couplings of the @xmath1 to a fermion @xmath78 are @xmath79 squaring the matrix element and integrating over the final state momenta for fixed @xmath65 and @xmath66 gives @xmath80 with @xmath81 where the sum is over fermion species .
note that @xmath82 as @xmath83 which is the result obtained by cutting a free propagator .
the variables @xmath8 , @xmath84 , @xmath85 , @xmath65 and @xmath86 in eq.([eq : diffxsec ] ) arise from products of standard covariants and external wave functions and therefore take real values dictated by the kinematics . | it is shown how to construct exactly gauge - invariant @xmath0-matrix elements for processes involving unstable gauge particles such as the @xmath1 boson .
the results are applied to derive a physically meaningful expression for the cross - section @xmath2 and thereby provide a solution to the long - standing _ problem of the unstable particle_. /#10= 0=0 1= 1=1 0>1 # 1 / |
toward the end of the millennium , cosmology has matured into a phenomenological science .
observational data now dominates aesthetics in the evaluation of cosmological models . of fundamental importance
is the determination of cosmological parameters , in particular , the hubble constant @xmath1 , the matter density faction @xmath2 , and the density fraction contributed by the cosmological constant @xmath3 .
observation of distant type ia supernovae ( sne ia ) has become an increasingly powerful means of measuring cosmological parameters ( @xcite , 1998 ; @xcite ; @xcite ) , because sne ia can be calibrated to be good standard candles at cosmological distances .
( @xcite ) a type ia sn is the thermonuclear explosion of a carbon - oxygen white dwarf in a binary when the rate of the mass transfer from the companion star is high .
the sn explosion blows the white dwarf completely apart .
the radioactive decay of the isotopes @xmath4ni and @xmath4co is responsible for much of the light emitted .
the sn lightcurve reaches a maximum about 15 days after the explosion and then declines slowly over years .
a sn can outshine the galaxy in which it lies .
two independent groups ( @xcite ) have made systematic searches for sne ia for the purpose of measuring cosmological parameters .
their preliminary results seem to indicate a low matter density universe , possibly with a sizable cosmological constant . even though the uncertainty in these results is large
, they clearly demonstrate that the observation of sne ia can potentially become a reliable probe of cosmology .
however , there are important systematic uncertainties of sne ia as standard candles , in particular , luminosity evolution and gravitational lensing . to constrain the evolution of sn ia peak absolute luminosities
, we need a large number of sne ia at significantly different redshifts ( low @xmath5 and @xmath6 ) , which is not available at present . both groups have assumed a smooth universe in their data analysis , although they include lensing in their error budgets . since we live in a clumpy universe
, the effect of gravitational lensing must be taken into account adequately for the proper interpretation of sn data . at present , the small number of observed high @xmath5 sne ia prevents adequate modeling of gravitational lensing effects . in this paper
, we propose a pencil beam survey of sne ia that could yield between dozens to hundreds of sne ia in redshift bins of 0.1 up to @xmath0 ( see 3 ) which would allow the proper modeling of gravitational lensing , as well as a quantitative understanding of luminosity evolution .
such a survey would yield a wealth of data which can be used to make accurate measurement of cosmological parameters and the sn rate , and provide powerful constraints on various aspects of the cosmological model . in 2 , we consider the weak lensing of sne ia .
we simulate sn ia luminosities by adding weak lensing noise and scatter in sn ia absolute magnitudes to standard candles placed at random redshifts .
we show how flux - averaging reduces the combined noise of gravitational lensing and scatter in sn ia absolute magnitudes . in 3 , we show that the number count of sne from a pencil beam survey is not sensitive to matter distribution in the universe ; it can be used as a test of models of cosmology and sn progenitors , or to measure the sn rate accurately . in 4 , we discuss the observational feasibility of the sn pencil beam survey .
5 contains conclusions .
in a sn ia hubble diagram , one must use distance - redshift relations to make theoretical predictions . unlike angular separations and flux densities ,
distances are not directly measurable , but they are indispensable theoretical intermediaries .
the distance - redshift relations depend on the distribution of matter in the universe . in a smooth friedmann - robertson - walker ( frw ) universe ,
the metric is given by @xmath7 $ ] , where @xmath8 is the cosmic scale factor , and @xmath9 is the global curvature parameter ( @xmath10 ) . the comoving distance @xmath11 is given by [ eq : r(z ) ] r(z)= \ { |_k|^1/2 _ 0^z dz^-1/2 } , where `` sinn '' is defined as sinh if @xmath12 , and sin if @xmath13 . if @xmath14 , the sinn and @xmath15 s disappear from eq.([eq : r(z ) ] ) , leaving only the integral .
the angular diameter distance is given by @xmath16 , and the luminosity distance is given by @xmath17 . however , our universe is clumpy rather than smooth . according to the focusing theorem in gravitational lens theory ,
if there is any shear or matter along a beam connecting a source to an observer , the angular diameter distance of the source from the observer is _ smaller _ than that which would occur if the source were seen through an empty , shear - free cone , provided the affine parameter distance ( defined such that its element equals the proper distance element at the observer ) is the same and the beam has not gone through a caustic .
an increase of shear or matter density along the beam decreases the angular diameter distance and consequently increases the observable flux for given @xmath5 .
( schneider , ehlers , & falco 1992 ) the observation of sne ia at @xmath18 is important for the determination of cosmological parameters , but the dispersion in sn ia luminosities due to gravitational lensing can become comparable to the intrinsic dispersion of sne ia absolute magnitudes because the optical depth for gravitational lensing increases with redshift .
if only a fraction @xmath19 ( known as the smoothness parameter ) of the matter density is smoothly distributed , the largest possible distance ( for given redshift ) for light bundles which have not passed through a caustic is given by the appropriate solution to the following equation : [ eq : dr ] g(z ) + _ m ( 1+z)^5 d_a=0 , where @xmath20 } $ ] .
( @xcite ) the @xmath21 form of eq.([eq : dr ] ) has been known as the dyer - roeder equation .
( @xcite ) fig.1(a ) shows magnitude versus redshift for the three cosmological models considered by riess et al .
( 1998 ) , scdm ( @xmath22 , @xmath21 ) , ocdm ( @xmath23 , @xmath21 ) , and @xmath24cdm ( @xmath23 , @xmath25 ) . for each cosmological model ,
the upper curve represents the completely clumpy universe ( empty beam , @xmath26 ) , while the lower curve represents the completely smooth universe ( filled beam , @xmath27 ) .
fig.1(b ) shows the same models relative to smooth ocdm ( filled beam , @xmath27 ) , the middle curve for each model now represents a universe with half of the matter smoothly distributed ( half - filled beam , @xmath28 ) .
clearly , at @xmath6 , there is degeneracy of distances in a flat clumpy universe and an open smooth universe , and also in an open clumpy universe and a flat smooth universe with a sizable cosmological constant , as has been noted by a number of previous authors .
( @xcite ) we can generalize the angular diameter distance @xmath29 by allowing the smoothness parameter @xmath19 to be _ direction dependent _ , i.e. , a property of the _ beam _ connecting the observer and the standard candle .
the smoothness parameter @xmath19 essentially represents the amount of matter that causes weak lensing of a given source .
since matter distribution in our universe is inhomogeneous , we can think of our universe as a mosaic of cones centered on the observer , each with a different value of @xmath19 .
this reinterpretation of @xmath19 implies that we have @xmath30 in regions of the universe in which there are above average amounts of matter which can cause magnification of a source .
( @xcite ) in order to derive a unique mapping between the distribution in distances and the distribution in the direction dependent smoothness parameter for given redshift @xmath5 , we _ define _ the direction dependent smoothness parameter @xmath19 to be the solution of eq.([eq : dr ] ) for given distance @xmath29 . at given redshift @xmath5 ,
the magnification of a source can be expressed in terms of the apparent brightness of the source @xmath31 , or in terms of the angular diameter distance to the source @xmath32 : [ eq : mu ] = = ^2 , where @xmath33 and @xmath34 are the flux of the source and angular diameter distance to the source in a completely smooth universe ( filled beam ) , and @xmath19 is the direction dependent smoothness parameter . since distances are not directly measurable , we should interpret eq.([eq : mu ] ) as defining a unique mapping between the magnification of a standard candle at redshift @xmath5 and the direction dependent smoothness parameter @xmath19 at @xmath5 ; @xmath19 parametrizes the direction dependent matter distribution in a well - defined manner . from the magnification distributions of standard candles at various redshifts , @xmath35 , with @xmath360.5 , 1 , 1.5 , 2 , 2.5 , 3 , 5 , found numerically by wambsganss et al . for @xmath37 , @xmath38 ( @xcite ) ,
wang ( 1999a ) has obtained simple empirical fitting formulae for the distribution of @xmath19 : [ eq : p(alpha ) ] p(|z)=c_norm , where @xmath39 , @xmath40 , @xmath41 , and @xmath42 depend on @xmath5 and are independent of @xmath19 .
they are given by [ eq : aq ] c_norm(z ) & = & 10 ^ -2 , + _
peak(z ) & = & 1.01350 -1.07857 ( ) + 2.05019 ( ) ^2 -2.14520 ( ) ^3 , + w(z ) & = & 0.06375 + 1.75355 ( ) - 4.99383 ( ) ^2 + 5.95852 ( ) ^3 , + q(z ) & = & 0.75045 + 1.85924 ( ) -2.91830 ( ) ^2 + 1.59266 ( ) ^3 .
@xmath43 is the normalization constant for given @xmath5 .
the parameter @xmath44 indicates the average smoothness of the universe at redshift @xmath5 , it increases with @xmath5 and approaches @xmath45 ( filled beam ) at @xmath46 ; the parameter @xmath47 indicates the width of the distribution in the direction dependent smoothness parameter @xmath19 , it decreases with @xmath5 .
the @xmath5 dependences of @xmath44 and @xmath47 are as expected because as we look back to earlier times , lines of sight become more filled in with matter , and the universe becomes smoother on the average .
the parameter @xmath48 indicates the deviation of @xmath49 from gaussianity ( which corresponds to @xmath50 ) .
models with different cosmological parameters should lead to somewhat different matter distributions @xmath49 . in the context of weak lensing of standard candles
, we expect the cosmological parameter dependence to enter primarily through the magnification @xmath51 to direction dependent smoothness parameter @xmath19 mapping at given @xmath5 ( the same @xmath19 corresponds to very different @xmath51 in different cosmologies )
. gravitational lensing noise in the hubble diagram can be reduced by appropriate flux averaging of sne ia in each redshift bin .
because of flux conservation , the average flux of a sufficient number of sne ia at the same @xmath5 from the same field should be the same as the true flux of the sne ia without gravitational lensing if the sample is complete .
it is convenient to compare the distance modulus of sne ia , @xmath52 , with the theoretical prediction _
0^p= 5 ( ) + 25 , where @xmath53 is the luminosity distance . before flux - averaging ,
we convert the distance modulus @xmath54 of sne ia into `` fluxes '' , @xmath55 .
we then obtain `` absolute luminosities '' , \{@xmath56 } , by removing the redshift dependence of the `` fluxes '' , i.e. , ( z_i ) = 4d_l^2(z_i|h_0,_m , _ )
f(z_i ) , where @xmath57 are the best - fit cosmological parameters derived from the unbinned data set \{@xmath58}. we then flux - average over the `` absolute luminosities '' \{@xmath59 } in each redshift bin .
the set of best - fit cosmological parameters derived from the binned data is applied to the unbinned data \{@xmath58 } to obtain a new set of `` absolute luminosities '' \{@xmath59 } , which is then flux - averaged in each redshift bin , and the new binned data is used to derive a new set of best - fit cosmological parameters .
this procedure is repeated until convergence is achieved .
this iteration should lead to the optimal removal of gravitational lensing noise and the accurate determinations of the cosmological parameters .
wang ( 1999b ) has applied this method to analyze the combined data from the two groups ( @xcite ) . to illustrate how flux - averaging can reduce the dispersion in sn ia luminosities caused by weak lensing ,
let us simulate the data by drawing @xmath60 random points from the redshift interval @xmath61 $ ] , each point represents a standard candle .
we add weak lensing noise by giving each standard candle a direction dependent smoothness parameter @xmath19 ( corresponding to @xmath62 ^ 2 $ ] ) drawn at random from the distribution @xmath49 given in 2.1 .
the scatter in sn ia absolute magnitudes , @xmath63 , can be written as m_abm = m_int+m_obs , where @xmath64 is the intrinsic scatter and @xmath65 is the observational noise .
we assume that both intrinsic scatter and observational noise are gaussian distributed in magnitude , with dispersions @xmath66 and @xmath67 respectively .
then the total scatter in sn ia absolute magnitudes is also gaussian distributed , i.e. , p(m_abm)= , with @xmath68 .
we take @xmath69 .
the absolute luminosity of each sn ia extracted from the data is ( z)&=&10 ^ -m_obs/2.5(|z)l_int(z ) , + & = & 10 ^ -m_obs/2.5(|z ) \ { 10 ^ -m_int/2.5 l(=1|z ) } + & = & l(=1|z)(|z ) 10 ^ -m_abm/2.5 .
we have used @xmath70 . for each sn
ia , the total noise is m = -2.5 ( ) .
let us average the fluxes of all sne ia in the redshift bin @xmath61 $ ] : = _
i=1^n_sn l(z_i ) .
the flux averaged noise is ( m ) _
-2.5 ( ) , where = .
table 1 lists the means and dispersions ( in the form of mean@xmath71dispersion ) of @xmath72 ( which are @xmath73 and @xmath74 ^ 2\rangle}\,$ ] ) , and @xmath75 ( which are @xmath76 and @xmath77 ^ 2\rangle}\,$ ] ) for various redshift bins with @xmath60=2 , 4 , and 9 sne ia in each bin , for @xmath78 random samples .
we have taken @xmath37 , @xmath38 .
cccccc @xmath5 & @xmath79 & @xmath60=2 & @xmath60=4 & @xmath60=9 + @xmath80 $ ] & 0.001 @xmath81 & -0.007@xmath82 & -0.012 @xmath83 & -0.015@xmath84 + @xmath85 $ ] & @xmath86 & -0.007 @xmath87 0.145 & -0.013 @xmath87 0.103 & -0.015 @xmath87 0.069 + @xmath88 $ ] & [email protected] & -0.006@xmath87 0.149 & -0.012 @xmath87 0.107 & -0.015 @xmath87
0.071 + @xmath90 $ ] & [email protected] & -0.006@xmath87 0.151 & -0.012 @xmath87 0.108 & -0.015 @xmath87
0.072 + @xmath91 $ ] & 0.005 @xmath92 & -0.005 @xmath870.155 & -0.012 @xmath87 0.111 & -0.015 @xmath87 0.074 + @xmath93 $ ] & 0.006 @xmath890.223 & [email protected] & [email protected] & [email protected] + @xmath94 $ ] & 0.007 @xmath890.228 & -0.004@xmath87 0.162 & [email protected] & -0.015@xmath87 0.078 + @xmath95 $ ] & 0.007 @xmath96 & [email protected] & [email protected] & -0.015 @xmath87 0.080 + @xmath97 $ ] & 0.008 @xmath870.238 & [email protected] & [email protected] & -0.015@xmath87 0.082 + @xmath98 $ ] & [email protected] & -0.004@xmath87 0.174 & -0.012@xmath87 0.125 & -0.016@xmath87 0.084 + the dispersion decreases roughly as @xmath100 .
the dispersion would reduce by 30% if the sample contains 2 sne ia ; 50% if the sample contains 4 sne ia .
even though gravitational lensing noise increases with redshift , the combined gravitational lensing and sn ia absolute magnitude scatter noise in the redshift interval @xmath101 $ ] can be reduced to the same level as in the absence of lensing by flux averaging over two sne ia .
note that the flux averaged luminosities are biased towards slightly higher luminosities .
this is as expected , because we have assumed that the intrinsic scatter and observational noise in the sn ia absolute magnitude are gaussian in _
magnitude_. it is straightforward to show that the mean of @xmath102 is @xmath103 $ ] , which corresponds to a bias of @xmath104 for @xmath105 .
if we assume that the intrinsic scatter and observational noise are gaussian in luminosity , the flux averaged luminosities become unbiased .
the sn number count is most sensitive to the sn rate as a function of @xmath5 , which depends on the specific rate of sne , as well as the number density of galaxies and their luminosity distribution .
the frequency of sne is a key parameter for describing the formation and the evolution of galaxies ; the winds driven by sne tune the energetics , and their production of metals determines the chemical evolution of galaxies and of clusters of galaxies ( @xcite ) .
the sn ii rate is related ( for a given initial mass function ) to the instantaneous stellar birthrate of massive stars because sne ii have short - lived progenitors ; the sn ia rate follow a slower evolutionary track , and can be used to probe the past history of star formation in galaxies .
accurate measurements of the sn rates at intermediate redshifts are important for understanding galaxy evolution , cosmic star formation rate and the nature of sn ia progenitors .
( @xcite ) the sn rates are very uncertain at present due to the small number of sne discovered in systematic searches .
( @xcite ) .
kolatt & bartelmann ( 1997 ) have estimated the sn ia average rate per proper time unit per comoving volume to be [ eq : snrate ] n_sn(z)=(100)^-1 ( h^-1)^-3 , a=0.0136 , b=0.067 , for @xmath106 .
changing @xmath107 to lower values leads to lower comoving densities but higher luminosities of galaxies at high @xmath5 ( @xcite ) .
eq.([eq : snrate ] ) has been derived assuming that all sne reside in galaxies , neglecting the redshift dependence of the specific sn rate ( number per unit luminosity per unit time ) , and using the number density of galaxies ( as function of redshift ) and the schechter function parameters derived from the canada france redshift survey ( @xcite ) , the apm survey ( @xcite ) , and the autofib survey ( @xcite ) .
we use eq.([eq : snrate ] ) for all cosmological models considered in this paper , because it is a very crude and conservative estimate , and we use it for the purpose of illustration only .
the expected number of sne ia in a field of angular area @xmath108 for an effective observation duration of @xmath109 up to redshift @xmath5 is n(z)&= & ^2 ^z_0 r^2(z ) , + & = & 22.4 ( ) ^2 ( ) _
0^zdz ^2 , where @xmath110 is the proper distance interval , @xmath11 is comoving distance , and the factor @xmath111 accounts for the cosmological time dilation .
note that the number counts fundamentally probe a different aspect of the global geometry of the universe than do the distance measures ( @xcite ) .
fig.2 shows the number of sne ia expected per 0.1 redshift interval as function of redshift for the same three cosmological models as in fig.1 , scdm ( @xmath22 , @xmath21 ) , ocdm ( @xmath23 , @xmath21 ) , and @xmath24cdm ( @xmath23 , @xmath25 ) . for a one square degree field , and an effective observation duration of one year , the total numbers of expected sne ia are 464 , 899 , and 1705 for scdm , ocdm , and @xmath24cdm respectively for @xmath5 up to 1.5 . the number of sne which are strongly lensed by galaxies is given by n_lensed(z)&= & ^z_0 dz ( z ) , + & = & 3.8110 ^ -2 ( ) ^2 ( ) ( ) + & & _ 0^zdz ^5 .
we have used the optical depth for gravitational lensing by galaxies ( @xcite ) ( z ) ^3 , where @xmath112 parametrizes the gravitational lensing effectiveness of galaxies ( as singular isothermal spheres ) .
fig.3 shows the number of strongly lensed sne as function of survey depth @xmath5 , for the same cosmological models as in fig.1 and fig.2 . for a one square degree field , and an effective observation duration of one year , the total numbers of strongly lensed sne ia are 0.2 , 0.6 , and 1.8 for scdm , ocdm , and @xmath24cdm respectively for @xmath5 up to 1.5 .
fig.2 and fig.3 show that the sn number counts from a pencil beam survey can be used to measure the sn rate at high redshifts and perhaps to probe cosmology .
because of the large number of sne in each redshift bin and the smallness of gravitational lensing optical depth , the sn number count should be insensitive to matter distribution in the universe , and should therefore provide a robust probe of the cosmological model .
note that the strongly lensed sne can be easily removed from the survey sample , because they appear as unusually bright sne .
the number of strongly lensed sne is very sensitive to the cosmological model and might be used to further constrain the cosmological model . but given the small number of strongly lensed sne ia expected in any realistic observational program , their usefulness may be limited . the sn number counts provide a combined measure of the cosmological parameters and the sn rate .
4 shows the parameter dependence of the sn ia number count per 0.1 redshift interval ( @xmath113 ) as function of redshift @xmath5 .
note that the dependences of the sn number count on the sn rate ( parametrized by @xmath114 and @xmath115 ) and the cosmological parameters are not degenerate and can be differentiated in principle . in practice , we are ignorant of the functional form of the sn rate as a function of @xmath5 . hence , we may apply the measurements of the cosmological parameters as priors to the sn number counts to obtain a direct measure of the sn rate in the universe for @xmath5 up to 1.5 , which can be used as a powerful constraint on the cosmological model . for a one square degree field , and an effective observation duration of one year , the sn rate per 0.1 redshift interval can be determined to 14 - 16% , 9 - 11% , 7 - 8% at @xmath116 for scdm , ocdm , and @xmath24cdm respectively .
a pencil beam survey would be efficient in the discovery of sne at @xmath117 through the combination of data from successive nights and the comparison of the latest frame of images with all previous frames .
sne ia are quite faint at @xmath118 , with ab magnitude in the i band of @xmath119 . because of the uv suppression due to line blanketing and the apparent ir suppression in the rest frame sn ia spectrum
, one should use a passband that corresponds to the wavelength range of 3000@xmath120 in the sn rest frame ; this means using the i , j , h , or k passbands to observe the @xmath6 sne .
sn searches from space ( hst and ngst ) are limited by small fields of view , a large scale sn search is only possible from the ground , where one is limited to the i band by the atmosphere . here
we discuss the observational feasibility of a pencil beam survey of sne ia up to @xmath118 in terms of exposure times in the i band ; although multiple band photometry should be obtained to constrain extinction and evolution of the sne . using values for the photometric parameters from the sloan digital sky survey ( 1 arcsec seeing , effective sky brightness of 20.3 mag / arcsec@xmath121 in the i band , seeing - dominated psf , etc ) , we find that the exposure time for a point source with ab magnitude of @xmath122 can be written as t = 13.94 hours ( ) ^2 ( ) ^2 10 ^ 0.8 ( i_ab-26 ) where @xmath123 is the signal - to - noise ratio , @xmath124 is the aperture of the telescope .
the above equation shows that the supernova pencil beam survey can be accomplished from a modest dedicated 4 meter telescope , which can image the same fields ( two pencil beams should be observed to keep the fields close to zenith ) every night , which can lead to the discovery of sne ia up to @xmath0 via appropriate combination of data from successive nights , and the light curves of sne ia at @xmath125 .
it should be feasible to monitor the faintest sne ia from the keck 10 meter telescope or the hst .
spectra of sne are required to determine whether they are type ia .
sne ia have a si absorption line at @xmath126 that may be used to identify the @xmath127 sne ia in ground - based observations .
the follow - up spectroscopy can be attempted on the keck low - resolution imaging spectrometer ( lris ) . assuming that we use the 300 grooves / mm grating which provide dispersions of 4.99@xmath128 per 48@xmath129 m ( 2 pixels on the ccd )
, we find the exposure time for 0.5 arcsec seeing to be t&=&28 hours ( ) ^2 ( ) 10 ^ 0.4 [ 2(i_ab-26)-(i^sky_ab-21 ) ] , + & = & 4.44 hours ( ) ^2 ( ) 10 ^ 0.4 [ 2(i_ab-25)-(i^sky_ab-21 ) ] , + & = & 12.33 hours ( ) ^2 ( ) 10 ^ 0.4 [ 2(i_ab-25)-(i^sky_ab-21 ) ] , where @xmath130 is the width of the slit , and @xmath131 is the sky brightness in the i band in units of mag / arcsec@xmath121 .
note that the si absorption feature at @xmath126 is not deep enough to be useful in a very noisy ( @xmath132 ) spectrum .
clearly , spectroscopic follow - up of the @xmath118 ( @xmath119 ) sne ia will require substantial observational resources ; nicmos or ngst will be more suitable for the spectroscopy of the sne ia at the highest redshifts discovered from the ground .
the keck lris can be used to obtain the spectra of the sne ia at more modest redshifts ( @xmath133 , @xmath134 ) .
a dedicated 4 meter telescope with a square degree field of view can be used to conduct other important scientific projects compatible with the sn pencil beam survey , such as qsos , kuiper belt objects , and in particular , weak lensing and the search of gamma - ray burst ( grb ) afterglows .
weak lensing is a powerful tool in mapping the mass distribution in the universe .
the large field of view and the depth of a pencil beam survey would be ideal for weak lensing measurements of field galaxies , which can be used to constrain the large scale structure in the universe .
grbs are perhaps the most energetic astrophysical events in the universe . currently , there are many competing theories to explain grbs .
grb afterglows contain valuable information , a statistically significant sample of grb afterglows can provide strong constraints on the grb theory . if beaming is involved in grbs , we expect most of the grb afterglows not to be associated with observable bursts .
schmidt et al . (
1998 ) have observed a few optical transients which were too short in duration to be sne ; these could be grb afterglows . since present observational data seem to indicate that grbs have associated host galaxies , the detection of host galaxies associated with short optical transients would support the interpretation of the latter as grb afterglows . since the grb host galaxies are typically fainter than @xmath135 , a pencil beam survey would be ideal in detecting these host galaxies of candidate grb afterglows .
we have proposed a pencil beam survey of sne ia which can yield from tens to hundreds of sne ia per 0.1 redshift interval for @xmath5 up to 1.5 , which would enable the quantitative consideration of the systematic uncertainties of sne ia as standard candles , in particular , luminosity evolution and gravitational lensing . using the perlmutter et al .
`` batch '' search technique repetitively over the same field in the sky , the pencil beam survey would be efficient in the discovery of sne at @xmath117 by allowing the comparison of the latest frame of images ( which may consist of combined data from successive nights ) with all previous frames .
the direct product of such a survey is the sn number count as a function of redshift ( see 3 ) , which is a combined measure of the cosmological parameters and the sn rate .
when the measurements of the cosmological parameters are applied as priors to the number count , we obtain a direct measure of the sn rate , which is a key parameter in the formation and evolution of galaxies .
the non - type - ia sne discovered by the pencil beam survey may be comparable to the type ia sne in number .
the most important and straightforward application of the data from the sn pencil beam survey is to reduce the gravitational lensing noise in a sn ia hubble diagram via flux averaging .
we have simulated sn ia luminosities by adding weak lensing noise ( using empirical fitting formulae given by wang 1999a ) and scatter in sn ia absolute magnitudes to standard candles placed at random redshifts .
we have shown that flux - averaging is powerful in reducing the combined noise of gravitational lensing and sn ia absolute magnitude scatter ( see 2 ) . because of the non - gaussian nature of the luminosity distribution of sne ia at given @xmath5 due to weak lensing
, the large number of sne ia in a given redshift interval at high @xmath5 is essential for the proper modeling and removal of the gravitational lensing effect .
the sn ia luminosity distribution in each redshift interval can be used to constrain the cosmological model ( in particular , the fraction of matter in compact objects ) by comparison with predictions of numerical ray - shooting .
( @xcite ) the completeness of the sn ia sample determines the effectiveness of the removal of gravitational lensing noise from the sn ia hubble diagram and the amount of information contained in the sn ia luminosity distribution in each redshift interval .
note that the magnitude limit of the survey can lead to observational bias against the most distant demagnified sne , therefore , the sne which are close to the magnitude limit of the survey should not be used to probe cosmology in the manner described in this paper . to ensure the maximum usefulness of the data
, scrupulous attention will have to be paid to photometric calibration , uniform treatment of nearby and distant samples , and an effective way to deal with reddening .
( @xcite ) note that although we can remove or reduce the effect of gravitational lensing on the sn ia hubble diagram , other systematics can affect the observed luminosity of sne ia .
for example , grey dust , an evolution of the reddening law , or evolution in the peak absolute luminosity of sne ia .
while it is possible to constrain grey dust and determine dust evolution through multi - band photometry at significantly different redshifts , the dimming with @xmath5 in peak absolute luminosity of sne ia is degenerate with the effect of low matter density .
( @xcite ) evolution will remain a caveat in the usage of sne ia as cosmological standard candles , unless one can somehow correct for the effect of evolution .
it is critical to obtain up to hundreds of sne ia at @xmath6 to constrain luminosity evolution , because we expect the luminosity evolution and low matter density to affect the distance modulus of sne ia through different functionals of @xmath5 , which should become distinguishable at @xmath6 .
we have proposed a survey cutoff of @xmath0 mainly for two reasons .
first , going to higher redshift makes obtaining the spectra of the sne ( which are needed to distinguish different types of sne ) practically impossible from the ground ; even obtaining the spectra of @xmath118 sne ia may prove impractical from the ground ( as it would require several clear nights per spectrum on the keck , see 4 ) .
observers have already demonstrated that sne ia up to @xmath136 can be found in the ongoing searches .
( @xcite ) at @xmath118 , the sne ia should be @xmath137 magnitudes fainter than at @xmath138 , it should be possible to discover them on the ground through deep imaging , and the follow - up spectroscopy can be done on nicmos or ngst ( see 4 ) .
second , predicted rest - frame sn ia rate per comoving volume as function of redshift seems to peak at @xmath138 .
( @xcite ) a pencil beam survey of sne up to @xmath0 will enable the accurate determination of the sn rate as function of redshift in the redshift region important for studying cosmic star formation rate and the sn ia progenitor models .
although the next generation space telescope can detect sne at as high redshifts as they exist ( @xcite ) , the estimated rate of detection is of order 20 sne ii per 4@xmath1394 arcmin@xmath121 field per year in the interval @xmath140 ( @xcite ) , and the detection rate of sne ia is likely smaller .
thus a ground - based pencil beam survey of sne is essential to complement the space - based sn searches .
the most challenging aspect of the sn pencil beam survey is obtaining spectra for the sne ia at redshifts close to 1.5 .
instead of waiting for future space equipments , we may find innovative ways of obtaining spectra from the ground .
the referee has pointed out that since we can find multiple sne at the same time on the same one square degree field ( the number of sne depends on cosmology ) , it may be possible to get multiple spectra at once with fibers .
the nominal numbers we have used for the proposed sn pencil beam survey , a one square degree field , a depth of @xmath0 , and an effective observation duration of one year ( which is equivalent to several years of actual observation ) , are optimistic but not implausible . the large sky coverage and
the long effective observation duration will probably require a large consortium of existing and new sn search teams through , for example , a dedicated 4 meter telescope which can be used at the same time for other important observational projects compatible with the sn pencil beam survey ( see 4 ) .
the goal of going up to @xmath0 in spectroscopy will require support from the keck lris and nicmos / ngst .
we conclude by noting that a sn pencil beam survey can yield enormous scientific return .
the observational efforts directed towards a sn pencil beam survey should be very rewarding .
it is a pleasure for me to thank joachim wambsganss for generously providing unpublished magnification distributions ; zeljko ivezic and todd tripp for explaining technical details concerning photometry and spectroscopy ; saul perlmutter for communicating details of the current supernova search by the supernova cosmology project ; ed turner for a careful reading of a draft of the manuscript and for helpful comments , the referee for encouraging and useful comments ; christophe alard , jim gunn , david hogg , rocky kolb , robert lupton , michael strauss , and tony tyson for helpful discussions .
aguirre , a.n . 1999 ,
astro - ph/9904319 carroll , s.m . , press , w.h . , &
turner , e.l .
1992 , ara&a , 30 , 499 drell , p.s . ;
loredo , t.j . ; & wasserman , i. 1999 , astro - ph/9905027 dyer , c.c . , & roeder , r.c .
1973 , , 180 , l31 .
ellis , r.s . , colless , m. , broadhurst , t. , heyl , j. , & glazebrook , k. 1996 , mnras , 280 , 235 ferrini , f. , & poggianti 1993 , , 410 , 44 frieman , j.a .
1997 , comments astrophys . , 18 , 323 , astro - ph/9608068 .
fukugita , m. , & turner , e.l .
1991 , mnras , 253 , 99 garnavich , p.m. et al .
1998 , , 493 , l53 goobar , a. & perlmutter , s. 1995 , , 450 , 14 holz , d.e . ,
& wald , r.m .
1998 , phys .
d , 58 , 063501 .
holz , d.e .
1998 , , 506 , l1 .
lilly , s.j . ,
tresse , l. , hammer , f. , crampton , d. , & le fvre , o. 1995 , , 455 , 108 .
linder , e.v .
1998 , , 497 , 28 loveday , j. , peterson , b.a . , efstathiou , g. , maddox , s.j .
1992 , , 390 , 338 kantowski , r. , vaughan , t. , & branch , d. 1995 , , 447 , 35 kantowski , r. 1998 , , 507 , 483 kolatt , t.s . , & bartelmann , m. 1998 , mnras , 296 , 763 madau , p. 1998 , in dodorico , s. , fontana , a. & giallongo , e. eds , pasp , the young universe : galaxy formation and evolution at intermediate and high redshift madau , p. 1998 , della valle , m. , & panagia , n. 1998 , mnras 297 , l17 metcalf , r.b . 1999 ,
mnras , 305 , 746 pain , r. et al .
1996 , , 473 , 356 perlmutter , s. , et al .
1997 , , 483 , 565 perlmutter , s. ; aldering , g. ; goldhaber , g. ; knop , r.a . ;
nugent , p. ; castro , p.g . ; deustua , s ; fabbro , s ; goobar , a ; groom , d.e . ; hook , i.m . ; kim , a.g . ; kim , m.y . ; lee , j.c . ; nunes , n.j . ;
pain , r. ; pennypacker , c.r . ; quimby , r. ; lidman , c. ; ellis , r.s . ; irwin , m. ; mcmahon , r.g . ; ruiz - lapuente , p. ; walton , n. ; schaefer , b. ; boyle , b.j . ; filippenko , a.v . ; matheson , t. ; fruchter , a.s .
; panagia , n. ; newberg , h.j.m . ; & couch , w.j .
1999 , , 517 , 565 renzini , a. , ciotti , l. , dercole , a. , & pellegrini , s. 1993 , , 419 , 52 riess , a.g . , press , w.h . , & kirshner , r.p .
1995 , , 438 , l17 riess , a.g . ;
filippenko , a.v . ; challis , p. ; clocchiatti , a. ; diercks , a. ; garnavich , p.m. ; gilliland , r.l . ; hogan , c.j . ; jha , s. ; kirshner , r.p . ; leibundgut , b. ; phillips , m.m . ; reiss , d. ; schmidt , b.p . ; schommer , r.a . ; smith , r.c .
; spyromilio , j. ; stubbs , c. ; suntzeff , n.b . ; & tonry , j 1998 , aj , 116 , 1009 riess , a.g . ,
et al . 1999 ,
astro - ph/9907037 ruiz - lapuente , p. & canal , r. 1998 , , 497 , l57 sadat , r. ; blanchard , a. ; guiderdoni , b. ; & silk , j. 1998 , astron .
astrophys . , 331 , l69 schneider , p. , ehlers , j. , & falco , e.e .
1992 , gravitational lenses , springer - verlag , berlin schmidt , b.p . ,
1998 , , 507 , 46 stockman , h.s .
; stiavelli , m. ; i m , m. ; & mather , j. 1998 , in smith , e.p . , & koratkar , a. eds , asp conf .
133 , science with the next generation space telescope , astron .
, san francisco turner , e.l .
1990 , , 365 , l43 van den bergh , s. , & mcclure , r.d .
1994 , , 425 , 205 wambsganss , j. , cen , r. , xu , g. , & ostriker , j.p .
1997 , , 475 , l81 wambsganss , j. 1999 , private communication .
wang , y. 1999a , astro - ph/9901212 , , in press wang , y. 1999b , astro - ph/9907405 , , accepted yungelson , l. & livio , m. 1998 , , 497 , 168 | type ia supernovae ( sne ia ) can be calibrated to be good standard candles at cosmological distances .
we propose a supernova pencil beam survey that could yield between dozens to hundreds of sne ia in redshift bins of 0.1 up to @xmath0 , which would compliment space based sn searches , and enable the proper consideration of the systematic uncertainties of sne ia as standard candles , in particular , luminosity evolution and gravitational lensing .
we simulate sne ia luminosities by adding weak lensing noise ( using empirical fitting formulae ) and scatter in sn ia absolute magnitudes to standard candles placed at random redshifts .
we show that flux - averaging is powerful in reducing the combined noise due to gravitational lensing and scatter in sn ia absolute magnitudes .
the sn number count is not sensitive to matter distribution in the universe ; it can be used to test models of cosmology or to measure the sn rate .
the sn pencil beam survey can yield a wealth of data which should enable accurate determination of the cosmological parameters and the sn rate , and provide valuable information on the formation and evolution of galaxies .
the sn pencil beam survey can be accomplished on a dedicated 4 meter telescope with a square degree field of view .
this telescope can be used to conduct other important observational projects compatible with the sn pencil beam survey , such as qsos , kuiper belt objects , and in particular , weak lensing measurements of field galaxies , and the search for gamma - ray burst afterglows . # 1#23.6pt astro - ph/9806185 +
september 10 , 1999 + to appear in apj , 531 , # 2 ( march 10 , 2000 ) |
buildings across the world contribute significantly to total energy usage .
figure [ fig : country_proportion ] shows the contribution of buildings to energy consumption in india , usa , china , korea and australia @xcite .
rapid rate construction of buildings presses the need to look into improving energy efficiency in buildings with a goal of decreasing overall energy footprint .
thus , towards the vision of sustainability in the age of dwindling natural resources , buildings need to be made more energy efficient .
measure twice and cut once"- so goes the old adage . with this adage in mind and
the goal of understanding building energy efficiency , collecting building energy data assumes prime importance .
traditionally , building energy data included * monthly * electricity bills collected * manually * by the utility companies and other * sporadic * data such as energy audits .
owing to the manual and sporadic nature of this collected data , the available data is very sparse and provides limited insights into building energy efficiency .
buoyed by the success of data sets such as mnist @xcite in instigating machine vision research , previous study @xcite suggests that research in building energy domain can be spurred by availability of data sets . while previously deemed improbable , collection of such data sets has become increasingly common due to low cost sensing devices enabling * high resolution * and * automated * data collection .
furthermore , governments and utilities across the world have started rolling out smart meters with an aim of developing a smart grid . beyond the envisioned applications by the government and the utilities ,
this data can also be used for developing an understanding into building energy .
smart meters typically report electricity consumption to the utility and to the end consumer at rates ranging from a reading every second to a reading every hour .
commercial buildings are also increasingly managed with building management systems ( bms ) which control different sub - systems such as lighting and hvac .
bms are computer based control systems for controlling and monitoring various building systems such as hvac and lighting and are typically used to manage commercial buildings .
bms sense several spatially distributed points across the building to monitor parameters required for control action .
buildings are also commonly equipped with ambient sensors for monitoring parameters such as light , temperature and humidity .
these ambient sensors are often coupled with security systems to raise intruder alarm or to maintain healthy ambient conditions via thermostat or bms control . with the advent of smart meters , increased usage of bms and ease of availability and installation of ambient sensors , there is a now a deluge of building energy data .
while collecting building energy data is easier than before , deployments can get increasingly hard to manage , especially when large number of sensors are deployed .
however , previous research @xcite has highlighted that building energy data can be obtained at different spatio - temporal granularities .
this highlights the need for * optimal instrumentation * that can have different connotations for spatial and temporal domain . for spatial domain ,
optimal instrumentation involves monitoring at a subset of locations while still being able to accurately predict at all the desired locations .
for the temporal dimension , optimal instrumentation involves sampling at a lower resolution while still being able to predict at a higher temporal resolution .
desired application may choose the optimal set of sensors considering cost - accuracy tradeoffs across these different granularities .
traditionally , there exist multiple building subsystems such as security , networking , hvac and lighting , each performing their own operations in isolation .
however , buildings are a unified ecosystem and optimal operations would require * interconnecting * these * sub - systems*. the combined information from the different sub - systems is greater than sum of individual information from each of the systems .
once the systems are interconnected , data coming from diverse systems can be used to improve upon the decision making for optimal building operations . *
inferred decision making * can help in identifying inefficiencies , raising alerts and suggest optimizations .
traditionally data from within the system has been used for simple decision making e.g. motion sensor based lighting control .
furthermore , utility companies previously relied on customers phone calls to detect power outages . however , utility companies can now leverage smart meter data from different homes to quickly detect power outages and plan accordingly .
temporal patterns in the electricity consumption data @xcite can be used to detect faulty operations , predict load profiles and optimize operations accordingly .
spatial patterns in the electricity consumption can identify different electrical sub - systems and study their effect on aggregate consumption .
the vast set of analysis possible further necessitates the importance of data sets collected from the real world which can be to used to simulate the effect of inferred decisions before performing optimizations .
buildings constitute ecosystems involving interactions between the occupants , the physical and the cyber world @xcite . the presence of occupants , each with their individual preferences makes the ecosystem even more complex .
traditionally , the control decisions for building operation are taken at a central facility level assuming certain desired operating conditions without involving the occupants in decisions regarding desired conditions .
such policies are bound to be energy inefficient and may cause occupant discomfort at times @xcite . moreover
, previous literature @xcite suggests that energy unaware occupant behavior can add upto one - third to a building s energy performance .
however , when empowered with actionable feedback , occupants may save upto 15% energy @xcite .
building occupants can also provide useful data such as comfortable temperature and light intensity levels , which can be used to optimally schedule the hvac systems .
thus , various optimizations in the building energy ecosystem can be enabled by * involving occupants*. all of optimal sensing from interconnected subsystems , resulting in decisions inferred from the rich dataset while involving the occupants will overall result in * intelligent operations*. such intelligent operations already exist but miss out one or more aspects discussed previously i.e. either they do not involve the occupants or are taken at a subsystem level without accounting for data from other subsystems in operation .
thus , based on literature in building energy domain , we have identified these five crust areas also called five is which are as follows : i ) instrument optimally ; ii ) interconnect sub - systems ; iii ) inferred decision making ; iv ) involve occupants and v ) intelligent operations .
figure [ fig : building_energy ] summarizes the relationship amongst the 5 is of building energy which span across broad fields such as sensor networks , behavioral sciences , data science and control science
. the interconnected nature of these 5 is makes the overall optimization of building energy efficiency a complex problem .
for solving such a complex problem of intelligent building operations , rich data from across the diverse systems , together with smart algorithms for data inference and participatory engagement of occupants is critical .
such rich datasets are now becoming a reality @xcite .
further , rich research exists in from recent times that address algorithms for data inference and understanding behavioral aspects of occupants feedback @xcite .
this work bring together these different aspects into a survey while pointing out the big opportunities that exist along each of the 5 is together with those existing at the system level .
having briefly discussed the various facets of data driven energy efficient buildings , we now discuss an example scenario covering all of these aspects .
we base our theme around the classic paper on pervasive computing @xcite .
john goes to bed at 11 pm in the night .
he sets the alarm on his smartphone at 5 am when he intends to go for a jog .
however , his health has not been at the best since the last few days affecting his sleep .
he decides to inform his alarm clock that it should wake him up at 5 am if he is able to sleep by 11:30 , else wake him up after he has completed 6 hours of his sleep .
john s smart surround sensors capture his sleep patterns and communicate to the alarm to ring at 6 am .
his sleep disturbances , coughing are also captured , archived and emailed to john for sending to his doctor .
john goes to the refrigerator to grab some cold water .
the refrigerator understands that the person who has come to pick up water is john .
john s heath system reminded his refrigerator that john has been advised to refrain from cold water .
john gets ready and leaves home for office at 9 am .
the security system , door and motion sensor register this movement .
this information is communicated to john s thermostat which starts ramping down .
his lights also turn off automatically sensing his absence .
john has a meeting at 10 am and thus he would only go to his cabinet to keep his bag .
based on the number of expected attendees , the thermostat in the meeting room starts ramping up , in order to achieve the desired temperature in time for the meeting .
john s washing machine at his home is scheduled to run for 1 hour before john returns in the evening .
his washing machine interacts with the grid and predicts the best time to run when the load on the grid is low is noon time when the electricity prices are low . around 3 pm
, winds carry away the clouds and there is bright sunshine .
john s solar system starts producing electricity .
part of this dc produce is directly fed to dc appliances in john s home and the surplus is stored in a battery .
john figured that he can save more money by locally consuming his generated solar energy as opposed to selling it to the grid . in the meanwhile , some of the artificial lights in john s office turn off in lieu of the sunlight available .
john gets an email from the nilm system installed at his home about his disaggregated monthly consumption .
the system recommends that the tungsten based lighting in his home is very inefficient and eating up 30% of his bill .
if john would replace the same with more efficient led based lighting , his overall bill would go down by 15% .
considering the cost of replacement , his roi period would be less than 6 months , after which he would be able to save 10 usd a month .
we now analyse the above scenario and see how each small piece in this giant picture is a reality today .
systems such as isleep @xcite have explored using smartphones and motion sensors for sleep quantity and quality detection . in order to wake up john after he has completed 6 hours of sleep
, his smartphone detects his sleep pattern .
his smartphone contains an app which detects his cough patterns @xcite and uploads the most critical data to his doctor .
when john wakes up and goes to the refrigerator , systems for energy apportionment @xcite ascertain that it is john and not someone else who is trying to draw cold water .
the refrigerator is connected to john s smartphone over ip and informs him that cold water could be injurious to his health .
several years of research in occupancy detection using variety of sensing modalities @xcite are able to detect that john s home is unoccupied when john leaves for his office .
smarter occupancy driven thermostats have been proposed in recent literature @xcite which control and save energy based on occupancy prediction and external temperature , when john leaves for his office . by interconnecting soft sensors @xcite such as office chat client and meeting software , the thermostat in john s personal cabinet does not ramp up since he has a meeting in the board room .
conference room management sensor system @xcite in the boardroom is alert to the number of occupants and drives the hvac accordingly .
meanwhile in john s home his washing machine turns on at 1 pm .
since this is an off - peak period , electricity is available at much cheaper rates at this time in comparison to the rest of the day .
john scheduled his washing machine to run for one hour when the electricity would be cheap , which was enabled by demand response strategy @xcite . due to availability of sunlight after 3 pm , dc appliances in john s home switch from the utility and consume raw dc power from the solar panel @xcite .
the lighting control system @xcite in john s office also senses the bright sunshine and saves power while maintaining comfortable lighting levels inside the office .
the smart meter installed at john s home is regularly collecting his electricity data and periodically sends john the disaggregated breakdown of power by appliances @xcite .
while many of these individual systems have been explored in the past in isolation and often in research settings , their application to the real world and their interconnection to complete the picture remains largely untested .
the stepping stone to making such complex scenario a reality is the high resolution multimodal data collected from each of these subsystems , interconnected together , with algorithms that can do efficient operations while accounting for john s preferences .
data is the new oil " .
data science has brought about a paradigm shift in the way problems are solved in myriad applications . from social networks to astronomy and subatomic physics , data has brought a new revolution enabling answering important questions .
we now briefly discuss how data can play a pivotal role towards the development of energy efficient building .
traditionally , building energy data was collected only for billing purposes .
figure [ fig : bill ] shows the electricity bills of residential apartments in london , uk and delhi , india .
usually , data for such bills is collected once a month manually by the utility companies .
the bill shown in figure [ fig : delhi ] only provides the units of energy consumed during the billing period .
commercial entities owing to their heavy consumption are often priced on a time of day based pricing .
thus , in addition to the total units consumed , commercial entities are also provided with units consumed in peak and non - peak hours . in both of these cases ,
the consumption information is coarse and provides limited actionable insights . with the advent of the smart meter and
some of the other sensors discussed in section [ sec : introduction ] , more building energy data is now available than ever before .
while the conventional process followed by utilities would generate a single reading per home per month , smart meters collecting data once every 15 minutes would be collecting 3000 times more data @xcite .
smart meters are often capable of collecting data at higher rates of once every minute which can amount to 240 tb of data collected from 5 million homes a month @xcite .
figure [ fig : big_data ] contrasts the volume of this smart meter data collected with other applications which have greatly benefited by the data deluge .
the availability of such building energy data enables one to answer several important questions , some of which are presented below : * consumers can be informed in real time about their consumption .
this may be seen analogous to the cell phone alerts we get after every data transaction or call made .
this may greatly help the end users to not only understand their usage but also act on it to conserve energy .
* utilities earlier used to rely on receiving phone calls from aggrieved customers for detecting outages . however , with smart meter data readily available , utilities can leverage it for speedier outage detection and allocate resources efficiently to reduce downtime .
* figure [ fig : bill_tod ] shows the time of day energy consumption of a building in iiit delhi campus .
when such detailed information is made available to the end users , they can take initiatives towards shifting their loads to normal or off - peak hours , when electricity is cheaper .
such measures also benefit the utility as the peak demand is reduced . *
figure [ fig : historical_bill ] shows the trend in energy consumption from a home in new delhi @xcite .
this decreasing energy usage can be attributed to decreasing temperatures and hence decreasing use of air conditioners .
developing an understanding into the correlations between weather and energy usage can allow better hvac control .
such information is present in the uk bill ( figure [ fig : london ] ) .
however , without detailed analysis the same does not translate into actionable savings . towards the realization of many of the above mentioned questions , building energy data requirements need to be categorized as per the application scenario .
we firstly discuss the different temporal and spatial resolutions at which such data can be collected .
* temporal resolution : * in accordance with the intended application , building energy data is collected at different temporal resolutions .
energy audits are performed once every few years @xcite and provide key insights into several aspects of buildings including energy efficiency and comfort levels .
such audits are costly and require significant instrumentation . at a lower resolution of once a month
utility companies collect total energy consumption .
as discussed previously this provides limited insights to the end user .
commercial entities often monitor their power factor daily as they are liable to be penalized if the power factor drops below a certain threshold .
most of the above considered data collection is largely manual and sporadic .
monitoring at higher resolutions of 1 hour or lesser requires automated mechanisms such as installation of smart meters . with the national smart meter roll outs across many countries as discussed in section [ sec : introduction ] and initiatives such as greenbutton , collecting and accessing electricity data
has become much easier for the end consumers .
certain applications such as non - intrusive load monitoring ( nilm ) @xcite may require high frequency data with sampling rates of more than a thousand samples every second @xcite .
we summarize the temporal variations in building energy data collection in table [ tab : temporal ] .
.temporal variations in building energy data collection [ cols="^,^",options="header " , ] we specifically discuss in detail about the well studied problem of non - intrusive load monitoring ( nilm ) and discuss how it spans across the 5 is . non - intrusive load monitoring ( nilm ) or energy disaggregation is the process of breaking down the energy measurement observed at a single point of sensing into constituent loads @xcite .
figure [ fig : nilm ] shows disaggregated consumption across a day for different appliances from the iawe @xcite data set . in 2006 , darby et al .
@xcite suggested that providing detailed electricity information feedback to end users can lead to 5 - 15% electricity usage reduction by behavioral change .
recently , chakravarty et al .
@xcite performed a study across more than 300 users in california and observed mean reduction of approx .
15% when disaggregated information and real time electricity information is provided to end users . with the advent of smart metering infrastructure ,
as discussed in section [ sec : introduction ] , nilm has had a lot of renewed interest .
a number of startups which provide itemized electrical usage to households over cloud based services have emerged recently .
the recent interest has also led to the release of data sets meant to aid nilm research . of these redd @xcite ,
blued @xcite , ampds @xcite , uk - dale @xcite , eco @xcite , sust - data @xcite , greend @xcite , combed @xcite , berds @xcite and pecan @xcite have been specifically released for nilm like applications .
apart from these data sets , other data sets such as tracebase @xcite , he s @xcite , smart * @xcite and iawe @xcite can be used for several other building applications , including nilm .
we now discuss the nilm problem across the five is .
* instrument optimally : * nilm involves breaking down the aggregate meter data measured at a single point into constituent appliances or sub - meters lower in the electrical tree .
thus , instrumentation for nilm involves measuring electrical parameters at both the aggregate and the sub - metered level ( desired for ground truth ) .
the aggregate power readings are typically measured using smart meters . in special cases , when high frequency data ( more than 1 khz ) is required , sophisticated data acquisition systems ( daqs )
sub - metered power data is typically measured either at the circuit level using current transformer ( ct ) based sensors or at appliance level using appliance sensors .
* interconnect sub - systems : * while in the classic nilm problem , interconnection across different data streams is not studied , some studies use the extra information from other modalities to improve disaggregation .
previous work @xcite indicate the correlation between energy usage and external temperature .
berges et al .
@xcite correlate occupancy sensors with electrical data to identify potential savings in unoccupied rooms .
similarly , other ambient sensors can also be interconnected with electrical sensors to obtain additional insights into the disaggregation problem @xcite .
* inferred decision making : * the vanilla use case of the nilm implementation provides an itemized breakdown of the electrical load .
this inference problem can be viewed as an inverse classification problem .
previous work has related this problem to source separation which is a well studied problem in sound processing .
hart et al .
@xcite , in their seminal work on nilm , proposed a simplistic combinatorial based and a simple edge detection based nilm algorithm .
both these algorithms form the foundation behind many of the state - of - the - art algorithms .
markovian analogues of combinatorial optimization formulation led to factorial hidden markov @xcite among other hidden markov models .
several nilm approaches have been proposed in the recent past @xcite and a rich overview has been captured in several recent work as well @xcite .
it must be pointed that most of the prior research includes * supervised * methods in a * centralized * * offline * setup .
* involve occupants : * as discussed earlier in this section , providing itemized feedback to end users has shown to reduce their end consumption .
thus , the occupants can be involved by not only providing them with itemized billing , but also providing actionable suggestions on top .
another interesting occupant involvement may arise from devising novel techniques for ground truth collection , which eliminates the need for appliance metering .
* intelligent operations : * real time control actions to save electricity based on disaggregated information require complicated interactions with the control systems .
current literature is thin on the aspect of automated control beyond the usual feedback that nilm systems provide . in this section
we briefly mention some of the challenges and opportunities in nilm research . * * need for extensive deployment : * for certifying an algorithm on a previously unseen home , appliance level data must be collected from that home .
* * computationally expensive approaches : * many approaches are computationally expensive and thus model only the high energy consuming appliances . owing to this , detailed information about low energy consuming appliances is often not made available .
the computationally intractable approaches also limit the application in the real world setting . *
* supervised methods require sub - metered data : * supervised nilm approaches train on the sub - metered data and create a model for each appliance .
this requires ground truth instrumentation , motivating the need for development of novel unsupervised learning mechanisms .
optimizing building energy usage remains an area of concern in light of dwindling natural resources . due to this concern , efforts are being concentrated to develop an understanding into buildings .
these efforts have led to a deluge of data coming from a variety of sensors .
this availability of data is changing the way we consume our electricity information . in this paper
we highlighted some of the applications enabled by this data availability , such as early power outage detection , peak load reduction , electricity consumption reduction .
based on our literature survey of research in building energy domain , we identified five crust areas enabling data centric energy efficient buildings : i ) instrument optimally ii ) interconnect sub - systems iii ) inferred decisions iv ) involve occupants and v)intelligent operations . across each of these five areas
we present the state - of - the - art , the core challenges and the opportunities in the field .
finally , we categorize different building energy applications as per these five is and discuss non - intrusive load monitoring , a well studied problem in building energy domain , in greater detail . | buildings across the world contribute significantly to the overall energy consumption and are thus stakeholders in grid operations . towards the development of a smart grid
, utilities and governments across the world are encouraging smart meter deployments .
high resolution ( often at every 15 minutes ) data from these smart meters can be used to understand and optimize energy consumptions in buildings .
in addition to smart meters , buildings are also increasingly managed with building management systems ( bms ) which control different sub - systems such as lighting and heating , ventilation , and air conditioning ( hvac ) . with the advent of these smart meters , increased usage of bms and easy availability and widespread installation of ambient sensors , there is a deluge of building energy data .
this data has been leveraged for a variety of applications such as demand response , appliance fault detection and optimizing hvac schedules . beyond the traditional use of such data sets
, they can be put to effective use towards making buildings smarter and hence driving every possible bit of energy efficiency .
effective use of this data entails several critical areas from sensing to decision making and participatory involvement of occupants . picking from wide literature in building energy efficiency
, we identify five crust areas ( also referred to as 5 is ) for realizing data driven energy efficiency in buildings : i ) instrument optimally ; ii ) interconnect sub - systems ; iii ) inferred decision making ; iv ) involve occupants and v ) intelligent operations .
we classify prior work as per these 5 is and discuss challenges , opportunities and applications across them . building upon these 5
is we discuss a well studied problem in building energy efficiency - non - intrusive load monitoring ( nilm ) and how research in this area spans across the 5 is .
= 1 |
for a family @xmath0 of subsets of a set @xmath7 , the _ intersection graph _
@xmath8 of @xmath3 is defined as the graph with vertex - set @xmath3 , in which two vertices are adjacent if and only if the corresponding sets have non - empty intersection . for a graph @xmath9 , the
_ chromatic number _ of @xmath9 , denoted @xmath10 , is the least number of colors needed in a proper coloring of @xmath9 ( a coloring such that any two adjacent vertices have distinct colors ) .
when talking about a proper coloring of a family @xmath3 of subsets of a given set , we implicitly refer to a proper coloring of the intersection graph of @xmath3 , thus the chromatic number @xmath1 is defined in a natural way .
the chromatic number of families of geometric objects in the plane have been extensively studied since the sixties @xcite .
since it is possible to construct sets of pairwise intersecting ( straight - line ) segments of any size , the chromatic number of sets of segments in the plane is unbounded in general .
however , erds conjectured that triangle - free intersection graphs of segments in the plane have bounded chromatic number ( see @xcite ) .
this was recently disproved @xcite .
the conjecture of erds initiated the study of the chromatic number of families of geometric objects in the plane as a function of their _ clique number _ , the maximum size of subsets of the family that pairwise intersect @xcite . in this paper , we consider families of geometric objects in the plane for which the chromatic number only depends on local properties of the families , such as the maximum number of objects containing a given point of the plane .
consider a set @xmath11 of pseudo - disks ( subsets of the plane which are homeomorphic to a closed disk ) such that the intersection of the interiors of any two pseudo - disks is empty .
let @xmath12 be the planar hypergraph with vertex set @xmath3 , in which the hyperedges are the maximal sets of pseudo - disks whose intersection is non - empty .
a proper coloring of @xmath3 is equivalent to a coloring of @xmath12 in which all the vertices of each hyperedge have distinct colors .
if every point is contained in at most @xmath4 pseudo - disks , borodin conjectured that there exists such a coloring of @xmath12 with at most @xmath13 colors @xcite .
it was recently proven that this conjecture holds asymptotically @xcite ( not only in the plane , but also on any fixed surface ) . as a consequence
, @xmath3 can be properly colored with @xmath14 colors .
it seems natural to investigate the same problem when pseudo - disks are replaced by _ pseudo - segments_. these are continuous injective functions from @xmath15 $ ] to @xmath16 and are usually referred to as _ strings_. consider a set @xmath17 of such strings .
we say that @xmath18 is _ touching _ if no pair of strings of @xmath18 cross , and that it is _ @xmath4-touching _ if furthermore at most @xmath4 strings can `` touch '' in any point of the plane , i.e. , any point of the plane is contained in at most @xmath4 strings ( see figure [ fig : strings ] for an example ) .
note that the family of all touching sets of strings contains all _ contact systems of strings _ , defined as sets of strings such that the interior of any two strings have empty intersection .
in other words , if @xmath19 is a contact point in the interior of a string @xmath20 , all the strings containing @xmath19 distinct from @xmath20 end at @xmath19 . in @xcite , hlinn studied contact system of strings such that all the strings ending at @xmath19 leave from the same side of @xmath20 .
such a representation is said to be _ one - sided _
( see figure [ fig : contact ] for an example ) .
it was proved in @xcite that if a contact system of strings is @xmath4-touching and every contact point is one - sided , then the strings can be colored with @xmath21 colors . in this paper ,
our aim is to study @xmath4-touching sets of strings in their full generality . observe that if @xmath18 is @xmath4-touching , @xmath4 might be much smaller than the maximum degree of @xmath18 .
however , based on the cases of pseudo - disks and contact system of strings , we conjectured the following in the conference version of this paper @xcite : [ conj : multi ] for some constant @xmath22 , any @xmath4-touching set of strings can be colored with @xmath23 colors .
this conjecture was subsequently proved by fox and pach @xcite , who showed that any @xmath4-touching set of strings can be colored with @xmath24 colors ( where @xmath25 is the base of the natural logarithm ) . in section [ sec : general ] , we show how to slightly improve their bound for small values of @xmath4 .
we also show that for any odd @xmath4 , the clique on @xmath26 vertices can be represented as a set of @xmath4-touching strings , so the best possible constant @xmath19 in conjecture [ conj : multi ] is between @xmath27 and @xmath28 . in section [ sec : intersect ] , we give improved bounds when any two strings can intersect a bounded number of times . in section [ sec : contact ] , we restrict ourselves to contact systems of strings where any two strings intersect at most once ( called @xmath29-intersecting , which were previouly studied by hlinn @xcite .
he asked whether there is a constant @xmath19 such that every one - sided @xmath29-intersecting @xmath4-touching contact system of strings is @xmath30-colorable .
we prove that they are @xmath31-colorable , and that every @xmath4-touching contact system of segments is @xmath32-colorable .
note that we do not need our contact systems to be one - sided ( but adding this assumption offers a minor improvement on the additive constants ) . before giving general bounds ,
let us first mention two classical families of touching strings for which coloring problems are well understood .
if a @xmath4-touching set of strings has the property that the interior of each string is disjoint from all the other strings , then each string can be thought of as an edge of some ( planar ) graph with maximum degree @xmath4 . by a classical theorem of shannon , the strings
can then be colored with @xmath33 colors .
if moreover , any two strings intersect at most once , then they can be colored with @xmath34 colors by a theorem of vizing ( even with @xmath4 colors whenever @xmath35 , using a more recent result of sanders and zhao @xcite ) . in this sense , all the problems considered in this article can be seen as a extension of edge - coloring of planar graphs .
an _ @xmath36-monotone string _ is a string such that every vertical line intersects it in at most one point .
alternatively , it can be defined as the curve of a continuous function from an interval of @xmath37 to @xmath37 .
sets of @xmath4-touching @xmath36-monotone strings are closely related to bar @xmath4-visibility graphs .
bar @xmath4-visibility graph _ is a graph whose vertex - set consists of horizontal segments in the plane ( bars ) , and two vertices are adjacent if _ and only if _ there is a vertical segment connecting the two corresponding bars , and intersecting no more than @xmath4 other bars .
it is not difficult to see that the graph of any set of @xmath4-touching @xmath36-monotone strings is a spanning subgraph of some bar @xmath38-visibility graph , while any bar @xmath38-visibility graph can be represented as a set of @xmath4-touching @xmath36-monotone strings . using this correspondence
, it directly follows from @xcite that @xmath4-touching @xmath36-monotone strings are @xmath39 colorable , and that the clique on @xmath40 vertices can be represented as a set of @xmath4-touching @xmath36-monotone strings .
if the left - most point of each @xmath36-monotone string intersects the vertical line @xmath41 , then it directly follows from @xcite that the strings can be colored with @xmath42 colors ( and a clique on @xmath42 vertices can be represented by @xmath4-touching @xmath36-monotone strings in this specific way ) .
before proving any result on the structure of sets of @xmath4-touching strings in general , we make the following observation : [ obs : pla ] the family of intersection graphs of 2-touching strings is exactly the class of planar graphs . the class of planar graphs being exactly the class of intersection graphs of 2-touching pseudo - disks ( see @xcite ) it is clear that planar graphs are intersection graphs of 2-touching strings ( by taking a connected subset of the boundaries of each pseudo - disk ) .
furthermore , every intersection graph of 2-touching strings is contained in the intersection graph of 2-touching pseudo - disks , and is thus planar .
indeed , it is easy given a set of 2-touching strings @xmath17 to draw a set of 2-touching pseudo - disks @xmath11 such that @xmath43 for every @xmath44 $ ] .
the following result was proved by fox and pach @xcite ( this shows conjecture [ conj : multi ] ) . in the following , @xmath25 is the base of the natural logarithm .
[ th : fox ] any @xmath4-touching set of strings is @xmath45-colorable .
their theorem is a direct consequence on the following bound on the number of edges in a graph represented by a set of @xmath4-touching strings . their proof is inspired by the probabilistic proof of the crossing lemma . [ lem : fox ] any graph represented by a set of @xmath46 @xmath4-touching strings has less than @xmath47 edges . when @xmath48 , their proof can easily be optimized to show that the number of edges is less than @xmath49 .
hence , every such graph has a vertex with degree less than 24 .
these graphs are thus 23-degenerate and have chromatic number at most 24 .
we now show how to modify their proof to slightly improve this bound .
[ th : k3 ] any 3-touching set of strings is 19-colorable .
this is a direct consequence of the following lemma .
[ lem : k3 ] any graph represented by a set of @xmath46 3-touching strings has at most @xmath50 edges . our proof follows the same lines as that of lemma [ lem : fox ] in @xcite .
let @xmath18 be a set of @xmath46 3-touching strings , and let @xmath51 be the number of edges in the corresponding intersection graph @xmath9 .
we can assume that for any point @xmath52 contained in three different strings , one of the strings is `` sandwiched '' between the other two ( see figure [ fig : transform ] , left ) .
for otherwise , we can apply the local transformation described in figure [ fig : sandwich ] and turn @xmath52 into three different points , each containing exactly two strings , without changing @xmath9 . ]
we now define two spanning subgraphs @xmath53 and @xmath54 of @xmath9 as follows .
two strings @xmath55 and @xmath19 are adjacent in @xmath53 if they intersect and for every intersection point @xmath52 of @xmath55 and @xmath19 , there exists a third string @xmath56 that is sandwiched between @xmath55 and @xmath19 . for every pair of strings @xmath55 and @xmath19 adjacent in @xmath53 , let @xmath57 be an arbitrarily chosen intersection point between @xmath55 and @xmath19 ( with some third string sandwiched between @xmath55 and @xmath19 ) .
two strings @xmath55 and @xmath19 are adjacent in @xmath54 if they are adjacent in @xmath9 but not in @xmath53 , i.e. , if there exists an intersection point @xmath58 of @xmath55 and @xmath19 such that either @xmath58 is not contained in another string or @xmath58 is contained in a third string @xmath56 that is not sandwiched between @xmath55 and @xmath19 . for @xmath59 , the edge - set of @xmath60
is denoted by @xmath61 , and the cardinality of @xmath61 is denoted by @xmath62 .
[ cl1 ] @xmath63 for each edge @xmath64 , assume that @xmath65 gives a charge of 1 to the string @xmath19 sandwiched between @xmath55 and @xmath56 at @xmath66 .
let @xmath67 be the set of neighbors of @xmath19 in @xmath54 .
the total charge @xmath68 received by @xmath19 is at most the number of pairs of vertices @xmath69 such that in @xmath70 , @xmath71 is a 2-contact point .
if we modify the set of strings intersecting @xmath19 so as to only preserve those 2-touching points ( all the other pairs of strings are made disjoint ) , we obtain a planar graph with vertex - set @xmath67 and with @xmath68 edges .
it follows that @xmath72 .
summing for all strings @xmath19 , we obtain that the total charge @xmath73 . since @xmath74 , we have @xmath63 , as claimed . .
[ fig : transform ] ] we are now ready to prove the lemma .
we select each string of @xmath18 uniformly at random with probability @xmath52 ( to be chosen later ) . in the subset of chosen strings ,
we slightly modify each 3-touching point as depicted in figure [ fig : transform ] in order to preserve all edges in @xmath54 .
let @xmath75 be the new set of strings .
for each edge @xmath76 , the probability that @xmath65 appears as an edge in @xmath75 is @xmath77 , and for each edge @xmath78 , the probability that @xmath65 appears as an edge in @xmath75 is at least the probability that @xmath55 and @xmath56 were both selected and the string @xmath19 sandwiched between @xmath55 and @xmath56 at @xmath66 was not selected , which is @xmath79 .
the expected number of vertices of the graph represented by @xmath75 is then @xmath80 and its expected number of edges is at least @xmath81 by observation [ obs : pla ] , the graph is planar and therefore , @xmath82 .
this can be rewritten as @xmath83 .
taking @xmath84 , we obtain @xmath85 . the ideas of lemma [ lem : k3 ] can be used to slightly improve the multiplicative constant in theorem [ th : fox ] for all @xmath4 . since our improvement is minor ( we obtain a bound of @xmath86 instead of @xmath87 ) , we omit the details .
we now show that the constant @xmath19 in conjecture [ conj : multi ] is at least @xmath88 . for every odd @xmath89 , @xmath90
, there exists a set of @xmath4-touching strings @xmath91 such that the strings of @xmath91 pairwise touch and such that @xmath92 .
thus @xmath93 .
consider @xmath46 touching strings @xmath94 that all intersect @xmath46 points @xmath95 in the same order ( see the set of bold strings in figure [ fig : exstring1 ] for an example when @xmath96 ) , and call this set of strings an _ @xmath46-braid_.
for some @xmath97 , take three @xmath98-braids @xmath99 , @xmath100 , @xmath101 , and for @xmath102 , connect each of the strings of @xmath103 to a different intersection point of @xmath104 ( with indices taken modulo 3 ) , while keeping the set of strings touching ( see figure [ fig : exstring1 ] ) .
we call this set of touching strings a _ @xmath105-sun_. observe that a @xmath105-sun contains @xmath106 strings ( the _ rays _ ) that pairwise intersect , and that each intersection point contains at most @xmath107 strings .
we now take three @xmath105-suns and paste their respective rays as depicted in figure [ fig : exstring2 ] .
we obtain a @xmath108-touching set containing @xmath109 pairwise intersecting strings .
hence we need at least @xmath110 colors in any proper coloring of this set of strings .
let @xmath18 be a @xmath4-touching set of strings .
the set @xmath18 is said to be _ @xmath111-intersecting _ if any two strings intersect in at most @xmath111 points .
we denote by @xmath112 the multigraph associated to @xmath18 : the vertices of @xmath112 are the strings of @xmath18 , and two strings with @xmath113 common points correspond to two vertices connected by @xmath113 edges in @xmath112 .
we prove the following result ( which only supersedes theorem [ th : fox ] for @xmath114 ) .
[ th : chro ] any @xmath4-touching set @xmath18 of @xmath111-intersecting strings can be properly colored with @xmath115 colors .
again , the proof is based on an upper bound on the number of edges of such graphs .
[ lem : sbsup ] if @xmath18 is a @xmath4-touching set of @xmath46 @xmath111-intersecting strings , then @xmath112 has less than @xmath116 edges . for a string @xmath20 ,
let @xmath117 denote the number of strings intersecting @xmath20 , and let @xmath46 denote the number of strings of @xmath18 .
let us denote by @xmath118 the number of strings containing an intersection point @xmath19 ( for any @xmath19 , @xmath119 by definition ) , and let @xmath120 denote the number of intersection points of @xmath18 . note that @xmath121 for all intersection points @xmath19 .
let us slightly modify @xmath18 in order to obtain a set @xmath75 of 2-touching and @xmath111-intersecting strings .
for that , repeat the following operation while there exists an intersection point @xmath19 with @xmath122 . consider a string @xmath123 such that all the other strings at @xmath19 are on the same side of @xmath20 ( such a string exists since strings do not cross each other ) .
move @xmath20 along another string @xmath124 to a new intersection point @xmath125 in such way that the strings remain touching ( see figure [ fig : transform ] for the case @xmath48 ) .
each intersection point @xmath19 in @xmath18 corresponds to a set @xmath126 of intersection points in @xmath75 .
let @xmath127 be the number of intersection points in @xmath75 . by construction
, each @xmath126 has size exactly @xmath128 , hence @xmath129 .
since the intersection graph of @xmath75 is a planar graph with @xmath46 vertices , it has at most @xmath130 edges . as @xmath75 is @xmath111-intersecting , @xmath131 .
as @xmath132 for any intersection point @xmath19 in @xmath18 , we have @xmath133 finally , since the number of edges of @xmath112 is @xmath134 which equals @xmath135 , we have that @xmath112 has less than @xmath136 edges . in particular , if a @xmath4-touching set @xmath18 of strings is such that any two strings intersect in at most one point , theorem [ th : chro ] yields a bound of @xmath137 for the chromatic number of @xmath18 .
we suspect that it is far from tight : [ conj : uni ] there is a constant @xmath22 , such that every @xmath4-touching set of 1-intersecting strings can be colored with @xmath138 colors . in the next section ,
we show that this conjecture holds for @xmath4-touching ( straight - line ) segments , a special case of 1-intersecting strings .
it is interesting to note that even though the bound for @xmath4-touching @xmath111-intersecting graphs in conjecture [ conj : multi ] and theorem [ th : fox ] does not depend on @xmath111 , the chromatic number of these graphs has some connection with @xmath111 : sets of strings with @xmath139 have chromatic number at most @xmath137 , whereas there exists sets of strings with large @xmath111 and chromatic number at least @xmath140 .
note that the constant @xmath19 in conjecture [ conj : uni ] is at least 4 .
figure [ fig : sbinf1 ] depicts a 3-touching set of seven 1-intersecting strings , in which any two strings intersect .
hence , this set requires seven colors .
however this construction does not extend to @xmath4-touching sets with @xmath141 , it might be that the constant is smaller for higher @xmath4 . in figure
[ fig : sbinf2 ] , the @xmath4-touching set @xmath142 contains @xmath143 non - crossing strings , and is such that any two strings intersect .
hence , @xmath143 colors are required in any proper coloring .
in this section , all the sets of strings we consider are 1-intersecting ( any two strings intersect in at most one point ) .
an interesting example of 1-intersecting set of strings is the family of non - crossing ( straight - line ) segments in the plane .
this family is also known as _ contact system of segments _ , and has been studied in @xcite , where the authors proved that any bipartite planar graph has a contact representation with horizontal and vertical segments . more generally , a _ contact system of strings _ is a family of strings such that the interiors of the strings are pairwise non - intersecting .
in other words , if @xmath19 is a contact point in the interior of a string @xmath20 , all the strings containing @xmath19 distinct from @xmath20 end at @xmath19 .
a contact point @xmath52 is a
_ peak _ if every string containing @xmath52 has an end at @xmath52 . otherwise , that is if @xmath52 is an interior point of a string @xmath20 and an end for all the other strings containing @xmath52 , @xmath52 is _
flat_. a flat contact point @xmath52 is _ one - sided _ if all the strings ending at @xmath52 are on the same side of the unique string whose interior contains @xmath52 .
a contact system of strings in which every flat contact point is one - sided is also said to be one - sided .
it was proven by ossona de mendez @xcite that the intersection graph of any one - sided 2- or 3-touching set of segments is planar .
note that as a 2-touching set of segments is always one - sided , it is also 4-colorable .
moreover , the author proved that it is np - complete to determine whether a 2-touching set of segments is 3-colorable . in @xcite
, hlinn studied the clique and chromatic numbers of one - sided @xmath4-touching contact systems of strings .
he proved that the maximal clique in this class is @xmath144 and that the graphs in this class are @xmath21-colorable .
he also asked the following : is there a constant @xmath19 such that if a contact system of strings is @xmath4-touching , 1-intersecting , and one - sided , then it is @xmath30-colorable ?
note that conjecture [ conj : uni ] would imply a positive answer to this question .
in the first part of this section , we prove that 1-intersecting and @xmath4-touching contact systems of strings are @xmath145-colorable . in the second part ,
we show that any @xmath4-touching contact system of segments is @xmath32-colorable .
note that we do not assume the contact systems to be one - sided ( but we also show that adding this assumption slightly improves the additive constants in our results ) .
assume that there is a counterexample , i.e. a 1-intersecting @xmath4-touching contact system @xmath18 of @xmath46 strings such that @xmath148 has @xmath150 edges .
we take @xmath46 minimal , and with respect to this , @xmath51 maximal .
observe that @xmath148 is connected , since otherwise by minimality of @xmath46 , each connected component would have average degree less than @xmath151 and then we would have @xmath152 , a contradiction .
we now prove that every string of @xmath18 has at least two contact points .
assume that some string @xmath20 has at most one contact point , then by the previous observation , it has a unique contact point @xmath52 , which coincides with one of its ends .
if every element of @xmath18 contains @xmath52 , then @xmath148 is a clique on at most @xmath153 vertices and has @xmath154 edges , a contradiction .
since @xmath148 is connected and is not a clique , it means that there is a string @xmath155 at distance 2 from @xmath20 in @xmath148 , i.e. , @xmath155 does not contain @xmath52 and touches some string @xmath156 containing @xmath52 .
take such a string @xmath155 in such a way that the portion of @xmath156 between @xmath52 and @xmath157 does not contain any contact point .
then it is easy to redraw @xmath20 as a string with ends @xmath52 and some interior point of @xmath155 close from @xmath157 ( such that @xmath20 contains no other intersection points than its two ends ) .
this contradicts the maximality of @xmath51 .
this shows that every string of @xmath18 has at least two contact points .
this also implies that the two ends of each string of @xmath18 are contact points ( if not , delete the portion of a string between a free end and its first contact point ) .
let @xmath158 be the plane graph whose vertices are the contact points of @xmath18 , whose edges link two contact points if and only if they are consecutive on a string of @xmath18 , and whose faces are the connected regions of @xmath159 .
let @xmath160 and @xmath161 be the number of contact points of exactly @xmath162 strings of @xmath18 that are respectively peaks and flat .
let us denote by @xmath19 the total number of contact points , and note that @xmath163 . by counting the number of ends of a string of @xmath18 in two different ways
, we obtain that : consider a one - sided flat contact point @xmath52 and let @xmath20 be the unique string such that @xmath52 is an interior point of @xmath20 .
if we draw a small open disk @xmath165 containing @xmath52 , a unique face @xmath166 of @xmath112 has the property that @xmath167 is incident to @xmath20 , and to no other string containing @xmath52 .
we denote this face @xmath166 of @xmath112 by @xmath168 . remark that since @xmath18 is 1-intersecting , any face @xmath166 of @xmath112 contains at least @xmath169 vertices ,
thus at least @xmath170 edges can be added to @xmath112 ( inside @xmath166 ) with the property that @xmath112 remains planar . hence in total , one can add as many edges to @xmath112 as the number of one - sided contact points , while keeping @xmath112 planar .
since every flat 2-contact point is one - sided , and every planar graph on @xmath19 vertices has at most @xmath171 edges , it follows that @xmath112 has at most @xmath172 edges . as a consequence
, we have : let us consider the linear program ( lp1 ) defined on variables @xmath160 and @xmath161 with values in @xmath176 such that the equation ( [ eq1 ] ) and the inequality ( [ eq2 ] ) are verified , and such that the value @xmath51 defined by ( [ eq3 ] ) is maximized . here
@xmath46 is considered to be a constant ( it is not a variable of the linear program ) .
note that the solution @xmath177 of this problem is clearly an upper bound of the number of edges of @xmath148 . if @xmath179 , take a small @xmath180 and replace @xmath181 by @xmath182 and @xmath183 by @xmath184 .
then ( [ eq1 ] ) remains valid , inequality ( [ eq2 ] ) still holds ( both sides are increased by @xmath185 ) , while ( [ eq3 ] ) is increased by @xmath186 .
if for some @xmath188 , @xmath189 , choose a small @xmath180 and replace ( i ) @xmath190 by @xmath191 ; ( ii ) @xmath192 by @xmath193 ; and ( iii ) @xmath194 by @xmath195 .
then ( [ eq1 ] ) remains valid , inequality ( [ eq2 ] ) still holds ( the left - hand side and the right - hand side remain unchanged ) , while the value of @xmath51 in ( [ eq3 ] ) is increased by @xmath196 .
the function @xmath197 is a ( concave ) parabola with @xmath198 ( recall that @xmath188 , so @xmath199 ) and @xmath200 , so it is positive in the interval @xmath201 $ ] . if for some @xmath204 , @xmath205 , choose a small @xmath180 and replace ( i ) @xmath206 by @xmath207 ; ( ii ) @xmath208 by @xmath209 ; and ( iii ) @xmath210 by @xmath211 .
then ( [ eq1 ] ) remains valid , inequality ( [ eq2 ] ) still holds ( the left - hand side and the right - hand side remain unchanged ) , while the value of @xmath51 in ( [ eq3 ] ) is increased by @xmath212 .
the function @xmath213 is a ( concave ) parabola with @xmath214 ( recall that @xmath204 , so @xmath215 ) and @xmath200 , so it is positive in the interval @xmath216 $ ] .
it follows from the previous claims that @xmath217 , so ( [ eq2 ] ) gives @xmath218
. therefore , @xmath219 by equation ( [ eq1 ] ) , @xmath220 which implies @xmath221 .
this can be rewritten as @xmath222 now , by equation ( [ eq3 ] ) , @xmath223 if the contact system we consider is one - sided , the argument we used for flat 2-intersection points while establishing ( [ eq2 ] ) in the previous proof works for all flat points , and inequality ( [ eq2 ] ) can be replaced by the stronger @xmath225 let ( lp2 ) be the new linear program .
we can repeat the same argument as above and prove that in some optimal solution of ( lp2 ) , we also have @xmath226 .
similar computations give that in this case the graph @xmath148 has less than @xmath227 edges . as a consequence
: we now show how to modify the proof of theorem [ th : contact ] to prove that @xmath4-touching contact systems of segments are @xmath32-colorable . for this
we will need an additional idea , based on the notion of _
stretchability_. a contact system of strings is _ stretchable _ if there is a homeomorphism that transforms it into a contact system of segments . a _
pseudo - line _ is the homeomorphic image of a straight line .
an _ arrangement of pseudo - lines _ is a set of pseudo - lines such that any two of them intersect at most once , and when they do they cross each other .
we say that a contact system of strings is _ extendible _ if there is an arrangement of pseudo - lines , such that each string of the contact system is contained in a distinct pseudo - line of the arrangement .
it was proved by de fraysseix and p. ossona de mendez @xcite that a contact system of strings is stretchable _ if and only if _ it is extendible .
the proof is similar to that of theorem [ th : contact ] .
we consider a counterexample @xmath18 consisting of @xmath46 @xmath4-touching segments such that @xmath148 has @xmath230 edges .
again , we take @xmath46 minimal , and with respect to this , @xmath51 maximal .
as before , we can assume that @xmath148 is connected and that the two ends of each segment are contact points .
we again consider the plane graph @xmath158 whose vertices are the contact points of @xmath18 , whose edges link two contact points if and only if they are consecutive on a segment of @xmath18 , and whose faces are the connected regions of @xmath159 .
let @xmath160 and @xmath161 be the number of contact points of exactly @xmath162 segments of @xmath18 that are respectively peaks and flat .
let @xmath231 and @xmath163 .
recall that any face @xmath166 of @xmath112 contains at least @xmath169 vertices ( where @xmath232 , defined in the proof of lemma [ lem : contact ] , is the number of flat contact points `` incident '' to @xmath166 ) , thus at least @xmath170 edges can be added to @xmath112 ( inside @xmath166 ) with the property that @xmath112 remains planar .
we now show that , moreover , the vertices corresponding to the peaks of @xmath18 all lie on the outerface of @xmath158 .
this directly implies that we can add @xmath233 edges to @xmath112 , while keeping @xmath112 planar ( as a consequence , @xmath158 has at most @xmath234 edges ) . indeed ,
if some peak @xmath36 of @xmath18 is not incident to the outerface , we choose a segment @xmath20 containing @xmath36 and prolong @xmath20 after @xmath36 until it hits some other segment @xmath235 ( note that since @xmath20 and @xmath235 are segments , they did not intersect previously ) .
let @xmath75 be the new contact system of segments . by the maximality of @xmath51 ,
the point @xmath236 is a contact point containing @xmath34 segments ( including @xmath20 and @xmath235 ) in @xmath75 . to correct this
, we do the following . by the result of de fraysseix and p. ossona de mendez @xcite ,
@xmath75 is extendible , so let @xmath237 be an arrangement of pseudo - lines extending @xmath75 ( i.e. such that every segment of @xmath75 is contained in a different pseudo - line of @xmath237 ) .
slightly perturb the pseudo - line @xmath2 containing @xmath20 around @xmath238 , so that it now intersects @xmath235 in some interior point of @xmath235 , very close from @xmath238 .
the new arrangement @xmath239 is still an arrangement of pseudo - lines , and it extends a contact system of segments @xmath240 obtained from @xmath75 by slightly moving the end of @xmath20 , starting from @xmath238 , along @xmath235 .
this new contact system @xmath240 contradicts the maximality of @xmath51 .
we consider the linear program ( lp3 ) defined on variables @xmath160 and @xmath161 with values in @xmath176 such that the equation ( [ eq1 ] ) and the inequality ( [ eq5 ] ) are verified , and such that the value @xmath51 defined by ( [ eq3 ] ) is maximized .
claim [ claim8 ] ( the optimal solutions of ( lp3 ) are such that @xmath242 ) and claim [ claim9 ] ( the optimal solutions of ( lp3 ) are such that @xmath187 , for every @xmath188 ) are still verified .
claim [ claim7 ] is now replaced by the following stronger claim : if for some @xmath244 , @xmath205 , choose a small @xmath180 and replace @xmath208 by @xmath245 , and @xmath192 by @xmath246 .
then ( [ eq1 ] ) remains valid , inequality ( [ eq5 ] ) still holds ( the left - hand side is decreased by @xmath247 and the right - hand side remains unchanged ) , while the value of @xmath51 in ( [ eq3 ] ) is increased by @xmath248 .
it follows from claims [ claim8 ] , [ claim9 ] , and [ claim11 ] that @xmath249 , so ( [ eq5 ] ) gives @xmath250 . by equation ( [ eq1 ] ) , @xmath251 , which implies @xmath252 .
this can be rewritten as @xmath253 now , by equation ( [ eq3 ] ) , @xmath254 the graph @xmath148 has less than @xmath255 edges , as desired .
it follows that intersection graphs of @xmath4-touching segments are @xmath256-degenerate and then @xmath32-colorable .
note that the @xmath256-degeneracy may not be tight for every @xmath4 .
indeed , we only know graphs that are not @xmath257-degenerate for @xmath258 . those graphs are obtained in the following way for @xmath259 : first consider the segments in a straight - line drawing of a planar triangulation with minimum degree 5 and maximum degree 6 , such that any degree five vertices is at distance at least two from the outerface , and such that any pair of degree five vertices are at distance at least three appart .
now let @xmath260 be the segments whose ends respectively touch a 5- and a 6-contact points .
those are the only segments that touch less than 10 other segments ( they only touch 9 of them ) . to make each of them touch one more segment , prolong successively the segments @xmath156 , @xmath155 and so on , by their end that is at the 6-contact point , until reaching another segment .
figure [ fig : segd ] depicts @xmath4-touching sets of segments requiring @xmath143 colors , for @xmath261 .
however it does not appear to be trivial to extend this construction for any @xmath262 .
note that figure [ fig : segd3 ] also shows that there are intersection graphs of 3-contact representations of segments ( with two - sided contact points ) that are not planar .
if the contact system of segments we consider is one - sided , we can prove that in a counterexample maximizing @xmath51 , all peaks are on the outerface .
therefore , in this case inequality ( [ eq4 ] ) in the proof of theorem [ th : contactos ] can be replaced by the stronger @xmath263 a. pawlik , j. kozik , t. krawczyk , m. laso , p. micek , w.t .
trotter , and b. walczak , _ triangle - free intersection graphs of line segments with large chromatic number _
, j. combin .
theory ser .
b * 105 * ( 2014 ) , 610 . | for a family of geometric objects in the plane @xmath0 , define @xmath1 as the least integer @xmath2 such that the elements of @xmath3 can be colored with @xmath2 colors , in such a way that any two intersecting objects have distinct colors .
when @xmath3 is a set of pseudo - disks that may only intersect on their boundaries , and such that any point of the plane is contained in at most @xmath4 pseudo - disks , it can be proven that @xmath5 since the problem is equivalent to cyclic coloring of plane graphs . in this paper
, we study the same problem when pseudo - disks are replaced by a family @xmath3 of pseudo - segments ( a.k.a .
strings ) that do not cross . in other words , any two strings of @xmath3 are only allowed to `` touch '' each other .
such a family is said to be @xmath4-touching if no point of the plane is contained in more than @xmath4 elements of @xmath3 .
we give bounds on @xmath1 as a function of @xmath4 , and in particular we show that @xmath4-touching segments can be colored with @xmath6 colors .
this partially answers a question of hlinn ( 1998 ) on the chromatic number of contact systems of strings . |
consider a spacetime described by a metric @xmath6 , and let @xmath0 be a given scalar field . from these variables , one can construct another metric on the same manifold defined by @xmath7 where @xmath8 $ ] and @xmath9 $ ] are arbitrary functions of the scalar field @xmath2 , and @xmath10 our aim in this paper would be study the relationship between geometric quantities associated with @xmath11 and @xmath6 .
however , before doing that , let us first give a brief motivation for considering such geometries . under what circumstances can such geometries arise , and what might be their use in physical theories ? over the past several decades , it has been realized ( although not widely appreciated ) that the geometry relevant for describing matter in presence of gravitational field could be related to the spacetime geometry in a non - trivial manner , going beyond the well known relation based on _ conformal transformations _ which have otherwise been ubiquitous in such contexts . a very transparent presentation of this argument can be found , for example , in bekenstein s paper @xcite .
he argued that if one takes _
finsler _ , rather than _ riemannian _ , geometry as more fundamental , and imposes some physically motivated constraints , a finsler geometry can be re - cast in terms of a riemannian geometry with a metric that is related to the spacetime metric by transformations which are non - conformal . since the transformation in ( [ eq : qmetric ] ) is qualitatively of such form ( although differs in detail from the ones in @xcite ) , we refer to such geometries as finsler_esque _ " , and shall be working throughout this paper in this restricted context .
a more recent reason for considering such geometries comes from semi - classical and quantum gravity .
in particular , it was recently shown @xcite that space(time)s with a minimal length scale might be described by an _ effective _ metric " of the form ( [ eq : qmetric ] ) , with the non - local synge world - function bi - scalar replacing the field @xmath0 .
we discuss this geometry in detail in section [ sec : min - length ] below .
_ summary _ : i outline a method which gives , in a convenient form , the expressions for ricci scalar associated with the metric @xmath12 in terms of geometric quantities associated with the level surfaces @xmath13 of @xmath0 . in section [ sec : extrinsic ] , i focus on the geometry of @xmath13 and find several relationships between intrinsic and extrinsic geometric quantities associated with @xmath13 as embedded in @xmath12 and @xmath14 . in section
[ sec : modified - riccis ] , i use these relations in the gauss - codazzi equation to re - construct the ricci scalar of @xmath11 ( see eqs . ( [ eq : mrs - final ] , [ eq : jcjd ] ) ) , and give several specific applications , including a recent result in which a non - local disformal coupling based on synge world function arises naturally in spacetimes with a minimal length . in section [ sec : riccis2
] , i recast the derived expression for the ricci scalar in a manner which elegantly highlights the ( well - known ) contribution of the conformal part , and the additional contribution due to the non - conformal part expressed in a transparent , geometric way .
finally , i end with brief concluding remarks in section [ sec : concluding - remarks ] .
_ notation _ : the signature is @xmath15 for lorentzian spaces .
i will also use the convenient notation @xmath16 which is handy when working in @xmath17 dimensions .
given the metric @xmath12 , the inverse metric is easily found to be @xmath18 where @xmath19 and @xmath20 . to begin with ,
introduce the vectors @xmath21 which are normalized with respect to @xmath12 .
of course , any characterization of @xmath13 in @xmath12 must be based on these vectors .
the metric determinants are related by @xmath22 which follows from the _ matrix determinant lemma _ : @xmath23 where @xmath24 is an invertible square matrix , and @xmath25 are column vectors ( of same dimension as @xmath24 ) . *
the first fundamental form * using the above relations , one can immediately deduce the following relation between the induced metrics , or the first fundamental forms , of @xmath13 in @xmath11 and @xmath6 .
@xmath26 that is , the _ induced _ geometries on @xmath13 in any two space(time)s related by eq .
( [ eq : qmetric ] ) are related by a _
transformation_. this observation will enormously simplify the evaluation of ricci scalar of @xmath12 in terms of quantities associated with @xmath14 .
* the intrinsic ricci scalar * the above result implies that intrinsic geometries of @xmath13 are conformal to each other , and since the conformal factor is @xmath27 $ ] which is constant on @xmath13 , one immediately obtains @xmath28 which is a simple rescaling , and , most importantly , independent of @xmath29 . * the second fundamental form
* our next aim would be relate the extrinsic geometries , or the second fundamental forms , of @xmath13 in the two metrics .
these are defined by @xmath30 where @xmath31 and @xmath32 are the acceleration vectors associated with vectors @xmath33 and @xmath34 respectively .
@xmath35 since @xmath36 , we need the relation between christoffel connections of the two metrics .
this is given by @xmath37 we only need @xmath38 , which is relatively straightforward to obtain using the identities proved in appendix [ app : chr ] , from which we obtain @xmath39 it immediately follows that @xmath40 putting everything together , and noticing that hypersurface orthogonality implies @xmath41 , we finally obtain @xmath42\end{aligned}\ ] ] and @xmath43\end{aligned}\ ] ] * summary * to summarize , we have derived the following relations between first and second fundamental forms of @xmath44constant @xmath45 \nn \\ \mk & = & \frac{1}{\sqrt{\ma-\mb } } \biggl [ k + d_1 \nabla_{\bm q } \ln \omega \biggl]\end{aligned}\ ] ] of special interest are two cases which are displayed in table [ tab : conformal - vs - disformal ] , corresponding to conformal and disformal transformations .
having the relationship between intrinsic and extrinsic geometrical properties of @xmath13 , we can now re - construct the point wise ricci scalar of @xmath12 by using the gauss - codazzi relation : @xmath46 where @xmath47 and @xmath48 .
our aim is to express all the quantities on the rhs of the above equation in terms of quantities associated with @xmath14 , thereby obtaining the transformation of @xmath49 .
this is easily done by using the results of the previous sections , which give the required relationships for quantities appearing in rhs above . at this stage
, it is useful to trade off the function @xmath29 in terms of a new function @xmath50 , defined by @xmath51 the significance of introducing @xmath50 will become clear as we go along . in particular , the set of metric transformations of the form ( [ eq : qmetric ] ) turn to have an extremely simple composition law in terms of the functions @xmath52 ; see appendix [ app : composition - law ]
. using expressions from previous section , one can derive the following relations after few ( long ) algebraic steps : @xmath53 \nn \\ \mk { } ^2 & = & \alpha \l [ k^2 + d_1 k \nabla_{\bm q } \ln \ma + d_1 ^ 2 \l ( \nabla_{\bm q } \ln \omega \r)^2 \r ] \nn \\ \mnabla_{\bm t } \mk & = & \alpha \nabla_{\bm q } k + \frac{1}{2 } \l ( k + d_1 \nabla_{\bm q } \ln \omega \r ) \nabla_{\bm q } \alpha \nn \\ \phantom{\mnabla_{\bm t } \mk } & & \phantom{\alpha \nabla_{\bm q } k } + d_1 \alpha \nabla_{\bm q } \nabla_{\bm q } \ln \omega \nn \\
\macc^i & = & \omega^{-2 } \nabla_i a^i \label{eq : aux - eqs}\end{aligned}\ ] ] where @xmath54 as above .
we now have everything needed to evaluate the rhs of eq .
( [ eq : modrs1 ] ) .
putting everything together , and using the ( easily proved ) identity @xmath55 we get @xmath56 where @xmath57 \nn \\ \vspace{.2 cm } & & \hspace{.55 cm } \ ; + \l ( k + d_1 \nabla_{\bm q } \ln \omega \r ) \times \nabla_{\bm q } \ln \alpha \omega^2 \nn \\ \nn \\ \mathcal{j}_d & = & 2 r_{ab } q^a q^b + k_{ab}^2 -k^2 - 2 \nabla_i a^i \nn \\ & = & \epsilon \l ( r - r_{\sigma } - 2 \nabla_i a^i \r ) \label{eq : jcjd}\end{aligned}\ ] ] which is the required expression . the crucial role played by the _ non - conformal _ term in the metric eq .
( [ eq : qmetric ] ) , characterized by @xmath58 , is immediately obvious from the above relation . in particular , @xmath59 in the above expression is multiplied by @xmath60 ( see eq .
( [ eq : alphadef ] ) ) . for conformally related metrics , @xmath61 , and therefore this term does not play any role in conformal transformations .
this observation is crucial , since it couples the non - conformal part of the metric @xmath12 to the object @xmath62 which has a very special structure !
+ the first application we give is based on the third relation in eqs .
( [ eq : aux - eqs ] ) .
consider the congruence of integral curves of @xmath33 , which need not necessarily be geodesics .
then , the expansion associated with these congruences is @xmath63 , and it s rate of change along the curves given by @xmath64 .
the corresponding quantities in metric @xmath12 are given by @xmath65 and @xmath66 , and their relationship , translated to variables more familiar from the raychaudhuri equation , is given by @xmath67 where we have replaced @xmath68 on the rhs with @xmath69 - the derivative along the curve .
( the special case of conformal transformation corresponds to @xmath70 . )
the above relation gives the modification to the raychaudhuri equation , and hence should be useful in studying the focussing and de - focussing behavior of geodesics in finsler - riemann spaces . if the origin of the scalar field @xmath0 is due to some fundamental physics associated with quantum gravitational effects , such a modification would be key to understand the effect of quantum gravity on spacetime singularities . in this context , it is also worth pointing out the relation between the symmetric , traceless part of @xmath71 , or the so - called shear - tensor , which happens to be much simpler that the relation between @xmath71 themselves . using again eqs .
( [ eq : aux - eqs ] ) , we get @xmath72 and hence @xmath73 it is particularly obvious from the above that the conformal ( @xmath70 ) and disformal ( @xmath74 ) cases correspond to completely different scaling of the shear tensor associated with @xmath0 foliations . a more general discussion on the propagation of expansion
, shear and vorticity in the context of finsler - riemann geometries can be found , for example , in @xcite . as mentioned at the end of previous section , for conformal transformations , @xmath70 , and
the expression for ricci scalar reduces to @xmath75_{\alpha \omega^2=1 } \nn \\ \nn \\ & = & \omega^{-2 } \l [ r - 2 d_1 \omega^{-1 } \square \omega - d_1 d_4 \omega^{-2 } ( \nabla \omega)^2 \r]\end{aligned}\ ] ] since @xmath76 in this case , and hence the second term in @xmath77 in eq . ( [ eq : jcjd ] ) vanishes .
we have therefore re - derived the well known expression for conformal transformation of the ricci scalar , in a manner which gives a geometric origin for the @xmath78 dependent terms .
for the special case of disformal transformations ) , while sometimes all such metrics are called disformal .
this is just a matter of terminology . ] , @xmath79 , and we obtain @xmath80_{\alpha = \omega^2 } \label{eq : rsdisformal}\end{aligned}\ ] ] since disformal modifications of spacetime geometry play an important role in several studies such as modified gravity , the above expression can provide considerable insight into construction of sensible action for such models . in a recent work @xcite , it was argued that a spacetime with a minimal length is endowed , under certain conditions , with precisely a disformal structure based on the bi - scalar of geodesic distance , @xmath4 , between spacetime events @xmath3 and @xmath5 , with @xmath81 it is a straight forward exercise to plug the above form for @xmath78 in eq .
( [ eq : rsdisformal ] ) and obtain eq .
( 14 ) of @xcite . _ equi - geodesic surfaces _ :
the key role here is played by the congruence of geodesics emanating from a fixed spacetime event @xmath5 , and the surface comprised of events @xmath3 lying at constant geodesic interval from @xmath5 , which we call as the _ equi - geodesic _ surface .
we give below several properties of the foliation based on such surfaces , and also sketch the derivation of eq .
( 14 ) in @xcite which was based on the analysis presented here .
let us start with the key geometric quantities associated with @xmath82constant surfaces with one of the events , say @xmath5 , is fixed ; see fig.[fig : geodesic - congruence ] .
since the affinely parametrized tangent vector to the geodesic connecting @xmath5 to @xmath3 is the normal to the @xmath13 , and given by @xcite @xmath83 the extrinsic curvature tensor of @xmath82constant surface , @xmath13 , is given by @xmath84 this particular foliation , formed out of points which are at a fixed geodesic interval from a given point , has many interesting characteristics , all deriving from the fact that the bi - tensor @xmath85 has a well know covariant taylor series expansion at @xmath3 near @xmath5 @xcite : @xmath86 where @xmath54 , @xmath87 , and @xmath88 is the numerical value of the geodesic distance between @xmath5 and @xmath3 .
therefore , we see that the extrinsic geometry of such a equi - geodesic foliation " is very special , and completely characterized by the _ tidal tensor _ @xmath87 .
in fact , the intrinsic and extrinsic curvatures can be characterized by a systematic taylor expansion around @xmath5 , given by @xmath89 where @xmath90 , @xmath91 , and @xmath92 .
the _ exact _ form of the ricci scalar ( in which the above taylor expansions can be plugged if needed ) turns out to be @xmath93 where @xmath94 and @xmath95 are the induced and extrinsic curvatures of @xmath82const .
surfaces in _ flat _ spacetime .
note that , in flat space(time ) , these surfaces are simply hyperboloids , and hence maximally symmetric @xmath96 spaces with positive or negative curvature ; see fig.[fig : geodesic - congruence](b ) .
let us point out some special features of the above expression , which are absent for a generic @xmath0 , being a consequence of ( i ) disformal nature of the coupling , ( ii ) the very specific form of @xmath97 which implies that @xmath13 corresponds to @xmath82 constant surfaces .
the first two terms on rhs mimic the relationship between @xmath14 and @xmath12 .
the form of the rhs clearly indicates an interplay between @xmath98 and @xmath99 riemann flat : the disformal character of the modification @xmath100 only couples to the curvature of the background spacetime , that is : _ flat space(time ) is immune to the disformal modification of the above form _
the geometrical structure of the above expression might hold the key to a generic study of behavior a disformal spacetime near curvature singularities of @xmath6 in terms of focusing and de - focussing of geodesics .
@xmath101 @xmath102 ccc + + conformal transformations & & disformal transformations + & & + @xmath103 & & @xmath104 + & & + + +
let @xmath105 $ ] denote the ricci scalar associated with metric @xmath6 , and define @xmath106 which measures deviation from the conformal case , for which @xmath107 and hence @xmath108 . then , recalling that @xmath109 + \epsilon \l ( k_{ab}^2 + k^2 \r ) + 2 \epsilon \nabla_{\bm q } k - 2 \epsilon \bm \nabla \bm \cdot \bm a \nn \end{aligned}\ ] ] is the intrinsic ricci scalar of the @xmath13 foliation embedded in @xmath6 , eq .
( [ eq : mrs - final ] ) can be cast in the following elegant form : @xmath110 \ ; & = & \ ; \underbrace { \l ( 1 + \md \r ) \ ; { \bf ric}\l[\ma \bm g \r ] } _ { \rm the~conformal~part } - \ ; \ ; \omega^{-2 } \underbrace{\l[\ ; \md \ ; \overbrace{\l ( r_{\sigma , \bm g } + 2 \epsilon \bm \nabla \bm \cdot \bm a \r)}^{\rm purely~{\it intrinsic } } - \ ; \epsilon \dot \md \ ; \overbrace{\l ( k + d_1 \nabla_{\bm q } \ln \omega \r)}^{\rm purely~{\it extrinsic } } \;\r]}_{{\rm contribution~of~the~ } \bm t \otimes \bm t { \rm ~term } } \nn \\ \nn \\ \nn \\ \ ; & = & \ ; \l ( 1 + \md \r ) \ ; { \bf ric}\l[\ma \bm g \r ] - \ ; \ ; \md \ ; \l ( r_{\sigma } + 2 \epsilon \bm \nabla \bm \cdot \bm a \r)_{{\ma \bm h } } + \ ; \epsilon \dot \md \ ; \omega^{-1 } k_{\sigma , { \ma \bm h } } \label{eq : rs - geom - struc } \end{aligned}\ ] ] where @xmath111 .
we believe that the above expression holds the key to the importance of the non - conformal term in finsler like spaces .
it brings in some important geometric quantities associated with the @xmath2 foliation , and drastically alters the behavior of the ricci scalar in a manner which pure conformal transformations can not .
the above fact was drastically brought to focus recently in the context of small scale structure of spacetime in presence of a minimal length @xcite , where the non - conformal part leads to some remarkable cancellations which drastically alters the low energy manifestation of minimal length effects , leaving a finite , @xmath112 , relic .
this suggests that the non - conformal transformations , if they arise due to quantum gravitational effects , might have non - trivial effect on physics at all scales .
besides the above , the geometric approach advocated here should also find immediate application in several other physical contexts ; we here mention a few , and also point out related generalizations . * relativistic mond * : the class of metrics we have considered here have been most prominent in the context of modified newtonian dynamics ( mond ) , where disformal transformations have played an important role
. the relativistic generalization of mond , proposed by bekenstein @xcite , involves a metric @xmath113 constructed from the background metric @xmath14 , a normalised timelike vector field @xmath114 , and a scalar field @xmath2 .
the structure of this modification has been extensively studied , specifically in the context of _ cosmology_. the geometric formalism presented here should be useful to investigate the characteristics of @xmath113 , in particular , for better understanding of the contribution of non - conformal part of the deformation as well structure of modified action for the theory .
this is already suggested by the discussion in section [ sec : riccis2 ] above .
doing so would require a generalization of the present analysis to the case when the vector field in question @xmath115 .
such a generalization should be straightforward , and we hope to address it in future work .
we here simply illustrate the power of our method in a simplified context in cosmology . consider the simplest ( yet most studied ) case of @xmath116 ( flat ) , @xmath117 , flrw metric @xmath14 in standard co - moving coordinates @xmath118 with scale factor @xmath119 . to study deformations which respect the background symmetry , we assume that @xmath78 and @xmath29 are functions of @xmath120 only . further , since conformal deformations are already well understood , we set @xmath121 .
in such a case , @xmath122 , and the relevant foliation is the one provided by @xmath123constant surfaces .
for such a foliation , @xmath124 and @xmath125 .
referring to eq .
( [ eq : rs - geom - struc ] ) , and noting that @xmath126 , we immediately get the modified ricci scalar to be @xmath127 \ ; & = & \ ; \frac { { \bf ric}\l [ g_{ab } \r]}{1-\mb } - \frac{3 h(t ) \dot \mb}{(1-\mb)^2 } \end{aligned}\ ] ] where @xmath128 is the hubble parameter . the non - flat case , @xmath129 , is also easy to obtain by noting that the metric @xmath130 ( being maximally symmetric ) immediately yields @xmath131 , which can then be plugged in eq .
( [ eq : rs - geom - struc ] ) along with other terms .
* lanczos - lovelock ( ll ) lagrangians * : these class of lagrangians have played an important role in generalizing and understanding various classical and semi - classical aspects of gravity over the past decade @xcite .
therefore , generalizing the calculation for ricci scalar @xmath132 presented here to generic ll lagrangians would provide another most natural extension of the analysis presented here . not only would this extend the study of disformal modifications of gravity to lanczos - lovelock models of gravity , but it would open up a completely new avenue of research into semi - classical aspects of conformal and non - conformal deformations of ll lagrangians . let us briefly elaborate on this point .
the ll lagrangians @xmath133 of order @xmath134 are built from @xmath134 copies of curvature tensor @xmath135 , while the surface term @xmath136 which makes the variational problem well defined is built from sum of products of @xmath137 and @xmath71 .
therefore , although finding the bulk term requires the knowledge of full modified riemann tensor ( which remains a formidable task ) , the surface term @xmath138 , since it involves only induced geometry @xmath130 , can be evaluated in a straightforward manner from the expressions given here .
since the surface term has played a very important role in semi - classical gravity , especially black hole thermodynamics , it s disformal transformation can lead to new insights beyond those obtained from purely conformal transformations .
we hope to take up this calculation in future work .
the mathematical expressions derived here , and their geometric properties , should be useful in several contexts in which an object like @xmath12 arises , either as a ` physical metric ' to which matter fields couple @xcite , or through some effective quantum gravitational model @xcite .
we hope that these expressions and the method of derivation would be applicable to a wider class of problems in which the relevant scalar ( or bi - scalar ) degree of freedom arises through some physical consideration , and couples non - conformally to the spacetime geometry .
moreover , since the ricci scalar forms the basis for setting up the standard gravitational action , the expression for @xmath49 in the form given in eq .
( [ eq : rs - geom - struc ] ) might provide insight into construction of the action involving @xmath14 and @xmath0 .
_ acknowledgements _ the author thanks iucaa , pune , where part of this work was done , for kind hospitality .
we start with the following relations which are straightforward to establish : @xmath139 t_b
t_c \label{eq : tmp1}\end{aligned}\ ] ] where @xmath140 .
using eq .
( [ eq : tmp1 ] ) in eq .
( [ eq : chr - rel ] ) , a few steps of algebra give @xmath141 where @xmath142 t_b t_c - 2 \mb k_{(bc ) } \nn\end{aligned}\ ] ] +
as an interesting aside , we note the following composition law for the transformations we are considering in this paper . using the variables @xmath143 , let us consider the following class of metrics , all defined on the same manifold . @xmath144 + then , noting that @xmath145 , it is easy to show that @xmath146 t^{(0)}_a t^{(0)}_b \nn \\\end{aligned}\ ] ] which immediately yields the following simple composition law @xmath147 or symbolically @xmath148 a. kouretsis , m. stathakopoulos , p. stavrinos , relativistic finsler geometry . math .
, 37 : 223229 ( 2014 ) [ arxiv:1301.7652 ] ; s. basilakos , a. kouretsis , e. saridakis , p. stavrinos , phys .
d * 88 * , 123510 ( 2013 ) [ arxiv:1311.5915 ] . | we consider metrics related to each other by functionals of a scalar field @xmath0 and it s gradient @xmath1 , and give transformations of some key geometric quantities associated with such metrics .
our analysis provides useful and elegant geometric insights into the roles of _ conformal _ and _ non - conformal _ metric deformations in terms of intrinsic and extrinsic geometry of @xmath2-foliations . as a special case
, we compare _ conformal _ and _ disformal _ transforms to highlight some non - trivial scaling differences .
we also study the geometry of _ equi - geodesic _ surfaces formed by points @xmath3 at constant geodesic distance @xmath4 from a fixed point @xmath5 , and apply our results to a specific disformal geometry based on @xmath4 which was recently shown to arise in the context of spacetime with a minimal length . 0.5 in |
we consider the classical problem of multiclass classification , where given an instance @xmath4 , the goal is to predict the most likely label @xmath5 , according to some unknown probability distribution .
a common general approach to multiclass learning is to reduce a multiclass problem to a set of binary classification problems @xcite .
this approach is composable with any binary learning algorithm , including online algorithms , bayesian algorithms , and even humans .
a key technique for analyzing reductions is _ regret analysis _ , which bounds the regret of the resulting multiclass classifier in terms of the average classification regret on the induced binary problems . here
( formally defined in section [ notation ] ) is the difference between the incurred loss and the smallest achievable loss on the problem , i.e. , excess loss due to suboptimal prediction .
the most commonly applied reduction is one - against - all , which creates a binary classification problem for each of the @xmath0 classes .
the classifier for class @xmath6 is trained to predict whether the label is @xmath6 or not ; predictions are then done by evaluating each binary classifier and randomizing over those that predict `` yes , '' or over all labels if all answers are `` no '' .
this simple reduction is _ inconsistent _ , in the sense that given optimal ( zero - regret ) binary classifiers , the reduction may not yield an optimal multiclass classifier in the presence of noise .
optimizing squared loss of the binary predictions instead of the @xmath7 loss makes the approach consistent , but the resulting multiclass regret scales as @xmath8 in the worst case , where @xmath9 is the average squared loss regret on the induced problems .
the probing reduction @xcite upper bounds @xmath9 by the average binary classification regret .
this composition gives a consistent reduction to binary classification , but it has a square root dependence on the binary regret ( which is undesirable as regrets are between 0 and 1 ) . the probabilistic error - correcting output code approach ( pecoc ) @xcite reduces @xmath0-class classification to learning @xmath10 regressors on the interval @xmath11 $ ] , creating @xmath10 binary examples per multiclass example at both training and test time , with a test time computation of @xmath12 .
the resulting multiclass regret is bounded by @xmath13 , removing the dependence on the number of classes @xmath0 .
when only a constant number of labels have non - zero probability given features , the computation can be reduced to @xmath14 per example @xcite .
this state of the problem raises several questions : 1 .
is there a consistent reduction from multiclass to binary classification that does not have a square root dependence on @xmath9 @xcite ?
for example , an average binary regret of just @xmath15 may imply a pecoc multiclass regret of @xmath16 .
2 . is there a consistent reduction that requires just @xmath1 computation , matching the information theoretic lower bound ?
+ the well - known @xmath1 tree reduction distinguishes between the labels using a balanced binary tree , with each non - leaf node predicting `` is the correct multiclass label to the left or not ? '' @xcite . as shown in section [ tree ] , this method is inconsistent .
3 . can the above be achieved with a reduction that only performs pairwise comparisons between classes ?
+ one fear associated with the pecoc approach is that it creates binary problems of the form `` what is the probability that the label is in a given random subset of labels ? , '' which may be hard to solve .
although this fear is addressed by regret analysis ( as the latter operates only on avoidable , excess loss ) , and is overstated in some cases @xcite , it is still of some concern , especially with larger values of @xmath0 .
the error - correcting tournament family presented here answers all of these questions in the affirmative .
it provides an exponentially faster in @xmath0 method for multiclass prediction with the resulting multiclass regret bounded by @xmath17 , where @xmath9 is the average binary regret ; and every binary classifier logically compares two distinct class labels .
the result is based on a basic observation that if a non - leaf node fails to predict its binary label , which may be unavoidable due to noise in the distribution , nodes between this node and the root should have no preference for class label prediction . utilizing this observation
, we construct a reduction , called the _ filter tree _ , which uses a @xmath1 computation per multiclass example at both training and test time , and
whose multiclass regret is bounded by @xmath18 times the average binary regret .
the decision process of a filter tree , viewed bottom up , can be viewed as a single - elimination tournament on a set of @xmath0 players . using multiple independent single - elimination tournaments is of no use as it does not affect the _ average _ regret of an adversary controlling the binary classifiers .
somewhat surprisingly , it is possible to have @xmath18 complete single - elimination tournaments between @xmath0 players in @xmath1 rounds , with no player playing twice in the same round . an _ error - correcting tournament _ , first pairs labels in such simultaneous single - elimination tournaments , followed by a final carefully weighted single - elimination tournament that decides among the @xmath18 winners of the first phase . as for the filter tree ,
test time evaluation can start at the root and proceed to a multiclass label with @xmath1 computation .
this construction is also useful for the problem of robust search , yielding the first algorithm which allows the adversary to err a constant fraction of the time in the `` full lie '' setting @xcite , where a comparator can missort any comparison .
previous work either applied to the `` half lie '' case where a comparator can fail to sort but can not actively missort @xcite , or to a `` full lie '' setting where an adversary has a fixed known bound on the number of lies @xcite or a fixed budget on the fraction of errors so far @xcite .
indeed , it might even appear impossible to have an algorithm robust to a constant fraction of full lie errors since an error can always be reserved for the last comparison .
repeating the last comparison @xmath1 times defeats this strategy .
the result here is also useful for the actual problem of tournament construction in games with real players .
our analysis does not assume that errors are _ i.i.d .
_ @xcite , or have known noise distributions @xcite or known outcome distributions given player skills @xcite . consequently , the tournaments we construct are robust against severe bias such as a biased referee or some forms of bribery and collusion .
furthermore , the tournaments we construct are shallow , requiring fewer rounds than @xmath19-elimination bracket tournaments , which do not satisfy the guarantee provided here . in an @xmath19-_elimination bracket tournament _ , bracket @xmath6 is a single - elimination tournament on all players except the winners of brackets @xmath20 .
after the bracket winners are determined , the player winning the last bracket @xmath19 plays the winner of bracket @xmath21 repeatedly until one player has suffered @xmath19 losses ( they start with @xmath21 and @xmath22 losses respectively ) .
the winner moves on to pair against the winner of bracket @xmath22 , and the process continues until only one player remains .
this method does not scale well to large @xmath19 , as the final elimination phase takes @xmath23 rounds . even for @xmath24 and @xmath25 ,
our constructions have smaller maximum depth than bracketed @xmath26-elimination .
to see that the bracketed @xmath19-elimination tournament does not satisfy our goal , note that the second - best player could defeat the first player in the first single elimination tournament , and then once more in the final elimination phase to win , implying that an adversary need control only two matches .
[ [ paper - overview ] ] paper overview + + + + + + + + + + + + + + we begin by defining the basic concepts and introducing some of the notation in section [ notation ] .
section [ tree ] shows that the simple divide - and - conquer tree approach is inconsistent , motivating the filter tree algorithm described in section [ s : algorithm ] ( which applies to more general cost - sensitive multiclass problems ) .
section [ s : analysis ] proves that the algorithm has the best possible computational dependence , and gives two upper bounds on the regret of the returned ( cost - sensitive ) multiclass classifier .
subsection [ s : experiments ] presents some experimental evidence that the filter tree is indeed a practical approach for multiclass classification .
section [ sec : multi - elimination ] presents the error - correcting tournament family parametrized by an integer @xmath27 , which controls the tradeoff between maximizing robustness ( @xmath19 large ) and minimizing depth ( @xmath19 small ) .
setting @xmath28 gives the filter tree , while @xmath29 gives a ( multiclass to binary ) regret ratio of @xmath30 with @xmath1 depth .
setting @xmath31 gives regret ratio of @xmath32 with depth @xmath10 .
the results here provide a nearly free generalization of earlier work @xcite in the robust search setting , to a more powerful adversary that can missort as well as fail to sort .
section [ sec : lb ] gives an algorithm independent lower bound of 2 on the regret ratio for large @xmath0 .
when the number of calls to a binary classifier is independent ( or nearly independent ) of the label predicted , we strengthen this lower bound to @xmath26 for large @xmath0 .
let @xmath33 be the underlying distribution over @xmath34 , where @xmath35 is some observable feature space and @xmath36 is the label space .
the _ error rate _ of a classifier @xmath37 on @xmath33 is given by @xmath38.\ ] ] the _ multiclass regret _ of @xmath39 on @xmath33 is defined as @xmath40 the algorithms here extend to the _ cost - sensitive _ case , where the underlying distribution @xmath33 is over @xmath41^{k}$ ] .
the _ expected cost _ of a classifier @xmath42 on @xmath33 is @xmath43.\ ] ] here @xmath44^{k}$ ] gives the cost of each of the @xmath0 choices for @xmath45 . as in the multiclass case ,
the _ cost - sensitive regret _ of @xmath39 on @xmath33 is defined as @xmath46
one standard approach for reducing multiclass learning to binary learning is to split the set of labels in half , learn a binary classifier to distinguish between the two subsets , and repeat recursively until each subset contains one label .
multiclass predictions are made by following a chain of classifications from the root down to the leaves .
this tree reduction transforms @xmath33 into a distribution @xmath47 over binary labeled examples by drawing a multiclass example @xmath48 from @xmath33 and a random non - leaf node @xmath6 , and outputting instance @xmath49 with label 1 if @xmath50 is in the left subtree of node @xmath6 , and 0 otherwise .
a binary classifier @xmath39 for this induced problem gives a multiclass classifier @xmath51 , via a chain of binary predictions starting from the root .
the following theorem gives an example of a multiclass problem such that even if we have an optimal classifier for the induced binary problem at each node , the tree reduction does not yield an optimal multiclass predictor . for all @xmath52 , for all binary trees over the labels , there exists a multiclass distribution @xmath33 such that @xmath53 for any @xmath54 .
find a node with one subset corresponding to two labels and the other subset corresponding to a single label .
( if the tree is perfectly balanced , simply let @xmath33 assign probability 0 to one of the labels . )
since we can freely rename labels without changing the underlying problem , let the first two labels be @xmath55 and @xmath56 , and the third label be @xmath26 .
fix any @xmath57 . choose @xmath33 with the property that labels @xmath55 and @xmath56 each have a @xmath58 chance of
being drawn given @xmath45 , and label @xmath26 is drawn with the remaining probability of @xmath59 . under this distribution ,
the fraction of examples for which label @xmath55 or @xmath56 is correct is @xmath60 , so any minimum error rate binary predictor must choose either label @xmath55 or label @xmath56 .
each of these choices has an error rate of @xmath61 .
the optimal multiclass predictor chooses label @xmath26 and suffers an error rate of @xmath62 , implying that the regret of the tree classifier based on the optimal binary classifier is @xmath63 , which is strictly greater than 0 as @xmath64 .
filter tree .
each node predicts whether the left or the right input label is more likely , conditioned on a given @xmath65 .
the root node predicts the best label for @xmath45 . ]
the filter tree algorithm , illustrated by figure [ fig : filter - tree ] , is equivalent to a single - elimination tournament on the set of labels , structured as a binary tree @xmath66 over the labels . in the first round ,
the labels are paired according to the lowest level of the tree , and a classifier is trained for each pair to predict which of the two labels is more likely .
( the labels that do nt have a pair in a given round , win that round for free . )
the winning labels from the first round are in turn paired in the second round , and a classifier is trained to predict whether the winner of one pair is more likely than the winner of the other .
the process of training classifiers to predict the best of a pair of winners from the previous round is repeated until the root classifier is trained .
the key trick in the training stage ( algorithm [ alg : filter ] ) is to form the right training set at each interior node .
we use @xmath67 to denote the subtree of @xmath66 rooted at node @xmath68 , and @xmath69 to denote the set of leaves in the tree @xmath66 . a training example for node @xmath68
is formed conditioned on the predictions of classifiers in the round before it .
thus the learned classifiers from the first level of the tree are used to `` filter '' the distribution over examples reaching the second level of the tree . given @xmath45 and classifiers at each node ,
every edge in @xmath66 is identified with a unique label .
the optimal decision at any non - leaf node is to choose the input edge ( label ) that is more likely according to the true conditional probability .
this can be done by using the outputs of classifiers in the round before it as a filter during the training process : for each observation , we set the label to 0 if the left parent s output matches the multiclass label , 1 if the right parent s output matches , and reject the example otherwise .
define @xmath70 if label @xmath50 is in the left subtree of node @xmath71 ; otherwise @xmath72 .
+ set @xmath73 add @xmath74 to @xmath75 let @xmath76 @xmath77 set @xmath73 [ line3 ] let @xmath78 and @xmath79 be the two classes input to @xmath68 @xmath80-.18 in [ line5 ] let @xmath76 @xmath81 algorithm [ alg : cs - filter - tree ] extends this idea to the cost - sensitive multiclass case where each choice has a different associated cost , as defined in section [ notation ] .
the algorithm relies upon an _ importance weighted _ binary learning algorithm ` learn ` , which takes examples of the form @xmath82 , where @xmath4 is a feature vector used for prediction , @xmath83 is a binary label , and @xmath84 is the importance any classifier pays if it does nt predict @xmath50 on @xmath45 . the importance weighted problem can be further reduced to binary classification using the costing reduction @xcite , which alters the underlying distribution using rejection sampling on the importances .
this is the reduction we use here .
the testing algorithm is the same for both multiclass and cost - sensitive variants , and is very simple : given a test example @xmath4 , we output the label @xmath50 such that every classifier on the path from the root to @xmath50 prefers @xmath50 .
before analyzing the regret of the algorithm , we note its computational characteristics . since the algorithm is a reduction , we count the computational complexity of the reduction itself , assuming that oracle calls take unit time .
algorithm [ alg : filter ] requires @xmath1 computation per multiclass example , by searching for the correct leaf in @xmath1 time , then filtering back toward the root .
this matches the information theoretic lower bound since simply reading one of @xmath0 labels requires @xmath85 bits .
algorithm [ alg : cs - filter - tree ] requires @xmath10 computation per cost - sensitive example , because there are @xmath86 nodes , each requiring constant computation per example .
since any method must read the @xmath0 costs , this bound is tight .
testing requires @xmath1 computation per example to descend a binary tree .
any method must write out @xmath87 bits to specify its prediction .
algorithm [ alg : cs - filter - tree ] transforms each cost - sensitive multiclass example ( line [ line3 ] ) into importance weighted binary labeled examples ( line [ line5 ] ) , one for every non - leaf node @xmath68 in the tree .
this process implicitly transforms the underlying distribution @xmath33 over cost - sensitive multiclass examples into a distribution @xmath88 over importance weighted binary examples at each @xmath68 .
we can further reduce from importance weighted binary classification to binary classification using the costing reduction @xcite , which alters each @xmath88 using rejection sampling on the importance weights .
this composition further transforms @xmath88 into a distribution @xmath89 over binary examples .
let @xmath90 be a classifier for the binary classification problem induced at node @xmath68 .
the relevant quantity is the _ average binary regret _ ,
@xmath91 where @xmath92 , and @xmath93 is the importance weight formed in line [ line5 ] of algorithm [ alg : cs - filter - tree ] ( the difference in cost between the two labels that node @xmath68 chooses between on @xmath45 ) .
this quantity , which is just the average _ importance weighted _ binary regret of @xmath90 on @xmath88 , is induced by the reduction ( algorithm [ alg : cs - filter - tree ] ) .
the core theorem below relates @xmath94 to the regret of the resulting cost - sensitive classifier @xmath51 on @xmath33 .
again , given a test example @xmath4 , the classifier @xmath51 returns the unique label @xmath50 such that every @xmath90 on the path from the root to @xmath50 prefers @xmath50 .
this type of analysis is similar to boosting : at each round @xmath68 , the booster creates an input distribution @xmath88 and calls a weak learning algorithm to obtain a classifier @xmath90 , which has some error rate on @xmath88 .
the distribution @xmath88 depends on the classifiers returned by the oracle in previous rounds .
the accuracy of the final classifier on the original distribution @xmath33 is analyzed in terms of these error rates .
[ main ] for all binary classifiers @xmath39 and all cost - sensitive multiclass distributions @xmath33
, @xmath95 where @xmath92 , and @xmath93 is the importance weight formed in line [ line5 ] of algorithm [ alg : cs - filter - tree ] ( the difference in cost between the two labels that node @xmath68 chooses between on @xmath45 ) . before proving the theorem , we state the corollary for multiclass classification .
[ cor : multi ] for all binary classifiers @xmath39 and multiclass distributions @xmath33 , @xmath96 where @xmath97 is the depth of the tree @xmath66 .
since all importance weights are either 0 or 1 , we do nt need to apply costing in the multiclass case .
the proof of the corollary given the theorem is simple since for any @xmath48 , the induced @xmath98 has at most one node per level with induced importance weight 1 ; all other importance weights are 0
. therefore , @xmath99 .
theorem [ bound ] provides an alternative bound for cost - sensitive classification .
it is the first known bound giving a worst - case dependence of less than @xmath0 . for all binary classifiers @xmath39 and all cost - sensitive @xmath0-class distributions @xmath33 , @xmath100 where @xmath51 and @xmath101 are as defined above .
[ bound ] a simple example in section [ example ] shows that this bound is essentially tight .
the proof of theorem [ main ] uses the following folk theorem from @xcite .
[ th : translate]_(translation theorem @xcite ) _ for any importance - weighted distribution @xmath102 , there exists a constant @xmath103 $ ] such that for any classifier @xmath39 , @xmath104 = \langle c \rangle { \mathbf{e}}_{(x , y , c)\sim { p}'}[{\mathbf{1}}(f(x)\neq y)],\ ] ] where @xmath105 . thus choosing @xmath39 to minimize the error rate under @xmath106 is equivalent to choosing @xmath39 to minimize the expected cost under @xmath102 .
the costing @xcite reduction uses rejection sampling according to the weights to draw examples from @xmath106 given examples drawn from @xmath102 .
the remainder of this section proves theorems [ main ] and [ bound ] .
* proof of theorem [ main ] : * it is sufficient to prove the claim for any @xmath4 because that implies that the result holds for all expectations over @xmath45 . conditioned on the value of @xmath45 ,
each label @xmath50 has a distribution over costs with an expected value of @xmath107 $ ] .
the zero regret cost - sensitive classifier predicts according to @xmath108 $ ] .
suppose that @xmath51 predicts @xmath109 on @xmath45 , inducing cost - sensitive regret @xmath110 - \min_y { \mathbf{e}}_{\vec{c } \sim d\mid x } [ c_y].\ ] ] first , we show that the sum over the binary problems of the importance weighted regret is at least @xmath111 , using induction starting at the leaves .
the induction hypothesis is that the sum of the regrets of importance - weighted binary classifiers in any subtree bounds the regret of the subtree output . for node @xmath68
, each importance weighted binary decision between class @xmath78 and class @xmath79 has an importance weighted regret which is either @xmath112 or @xmath113 | = |{\mathbf{e}}_{\vec{c } \sim d|x } [ c_a ] - { \mathbf{e}}_{\vec{c } \sim d|x } [ c_b ] | , $ ] depending on whether the prediction is correct or not .
assume without loss of generality that the predictor outputs class @xmath79 .
the regret of the subtree @xmath67 rooted at @xmath68 is given by @xmath114 - \min_{y \in { l}(t_n ) } { \mathbf{e}}_{\vec{c } \sim d|x } [ c_y].\ ] ] as a base case , the inductive hypothesis is trivially satisfied for trees with one label .
inductively , assume that @xmath115 and @xmath116 for the left subtree @xmath117 of @xmath68 ( providing @xmath78 ) and the right subtree @xmath118 ( providing @xmath79 ) .
there are two possibilities .
either the minimizer comes from the leaves of @xmath117 or the leaves of @xmath118 .
the second possibility is easy since we have @xmath119 - \min_{y \in { l}(r ) } { \mathbf{e}}_{\vec{c } \sim d|x } [ c_y ] = r_r \leq \sum_{n ' \in r } r_{n ' } \leq \sum_{n ' \in { t_n } } r_{n ' } , \end{aligned}\ ] ] proving the induction .
for the first possibility , @xmath119 - \min_{y \in { l}(l ) } { \mathbf{e}}_{\vec{c } \sim d|x } [ c_y ] \\ & = { \mathbf{e}}_{\vec{c } \sim d|x } [ c_b ] - { \mathbf{e}}_{\vec{c } \sim d|x } [ c_a ] + { \mathbf{e}}_{\vec{c } \sim d|x } [ c_a ] - \min_{y \in { l}(l ) } { \mathbf{e}}_{\vec{c } \sim d|x } [ c_y ] \\ & = { \mathbf{e}}_{\vec{c } \sim d|x } [ c_b ] - { \mathbf{e}}_{\vec{c } \sim d|x } [ c_a ] + r_l \\ & \leq r_n + \sum_{n ' \in l } r_{n ' } \leq \sum_{n ' \in { t_n } } r_{n'},\end{aligned}\ ] ] which completes the induction .
the inductive hypothesis for the root is that @xmath120 . using the folk theorem from @xcite ( theorem [ th : translate ] in this paper ) , each @xmath121 is bounded by @xmath122 plugging this in and using definition ( [ abr ] ) , we get the theorem .
the proof of theorem [ bound ] makes use of the following lemma .
consider a filter tree @xmath66 on @xmath0 labels , evaluated on a cost - sensitive multiclass example with cost vector @xmath123^k$ ] .
let @xmath124 be the sum of importances over all nodes in @xmath66 , and @xmath125 be the sum of importances over the nodes where the class with the larger cost was selected for the next round .
let @xmath126 denote the cost of the winner chosen by @xmath66 . for any @xmath123^k$ ] ,
[ claim1 ] the inequality follows by induction , the result being immediate when @xmath128 .
assume that the claim holds for the two subtrees , @xmath117 and @xmath118 , providing their respective inputs @xmath129 and @xmath9 to the root of @xmath66 , and @xmath66 outputs @xmath9 without loss of generality . using the inductive hypotheses for @xmath117 and @xmath118 , we get @xmath130 . if @xmath131 , we have @xmath132 , and @xmath133 as desired .
if @xmath134 , we have @xmath135 and @xmath136 , completing the proof .
* proof of theorem [ bound ] : * fix @xmath137^k$ ] and take the expectation over the draw of @xmath98 from @xmath33 as the last step .
consider a filter tree @xmath66 evaluated on @xmath98 using a given binary classifier @xmath39 .
as before , let @xmath124 be the sum of importances over all nodes in @xmath66 , and @xmath125 be the sum of importances over the nodes where @xmath39 made a mistake . recall that the regret of @xmath66 on @xmath98 , denoted in the proof by @xmath138 , is the difference between the cost of the tree s output and the smallest cost @xmath139 .
the importance - weighted binary regret of @xmath39 on @xmath98 is simply @xmath140 .
since the expected importance is upper bounded by 1 , @xmath140 also bounds the binary regret of @xmath39 .
the inequality we need to prove is @xmath141 .
the proof is by induction on @xmath0 , the result being trivial if @xmath128 .
assume that the assertion holds for the two subtrees , @xmath117 and @xmath118 , providing their respective inputs @xmath129 and @xmath9 to the root of @xmath66 .
( the number of classes in @xmath117 and @xmath118 can be taken to be even , by splitting the odd class into two classes with the same cost as the split class , which has no effect on the quantities in the theorem statement . )
let the best cost @xmath139 be in the left subtree @xmath117 .
suppose first ( * case 1 * ) that @xmath66 chooses @xmath9 and @xmath142 .
let @xmath143 .
we have @xmath144 and @xmath145 .
the left hand side of the inequality is thus @xmath146 the first inequality follows from lemma [ claim1 ] .
the second and fourth follow from @xmath147 .
the third follows from @xmath148 .
the last follows from @xmath149 for @xmath150 .
the proofs for the remaining three cases ( @xmath151 , @xmath152 , and @xmath153 ) use the same machinery as the proof above .
* case 2 * : @xmath66 outputs @xmath129 , and @xmath154 . in this case
the left hand side can be rewritten as @xmath156 the first inequality follows from the lemma , the second from @xmath157 , the third from @xmath148 , the fourth from @xmath158 , and the fifth because @xmath135 .
* case 3 * : @xmath66 outputs @xmath129 , and @xmath159 .
we have @xmath155 .
the left hand side can be written as @xmath160 the first inequality follows from the inductive hypothesis and the lemma , the second from @xmath161 and @xmath162 , and the third from @xmath163 and @xmath164 .
* case 4 * : @xmath66 outputs @xmath9 , and @xmath159 .
let @xmath165 .
we have @xmath166 .
the left hand side can be written as @xmath167 the first inequality follows from the inductive hypothesis and the lemma , the second from @xmath148 , and the third from @xmath168 .
the last three terms are upper bounded by @xmath169 and thus can be ignored , yielding @xmath170 , which completes the proof .
taking the expectation over @xmath98 completes the proof .
the following simple example shows that the theorem is essentially tight .
let @xmath0 be a power of two , and let every label have cost 0 if it is is even , and 1 otherwise .
the tree structure is a complete binary tree of depth @xmath18 with the nodes being paired in the order of their labels .
suppose that all pairwise classifications are correct , except that class @xmath0 wins all its @xmath18 games leading to cost - sensitive multiclass regret 1 .
we have @xmath171 , @xmath172 , and @xmath173 , leading to the regret ratio @xmath174 almost matching the theorem s bound of @xmath175 on the ratio .
there is a variant of the filter tree algorithm , which has a significant difference in performance in practice .
every classification at any node @xmath68 is essentially between two labels computed at test time , implying that we could simply learn one classifier for every pair of labels that could reach @xmath68 at test time .
( note that a given pair of labels can be compared only at a single node , namely their least common ancestor in the tree . )
the conditioning process and the tree structure gives us a better analysis than is achievable with the all - pairs approach @xcite .
this variant uses more computation and requires more data but often maximizes performance when the form of the classifier is constrained .
we compared the performance of filter tree and its all - pairs variant described above to the performance of all - pairs and the tree reduction , on a number of publicly available multiclass datasets @xcite .
some datasets came with a standard training / test split : ` isolet ` ( isolated letter speech recognition ) , ` optdigits ` ( optical handwritten digit recognition ) , ` pendigits ` ( pen - based handwritten digit recognition ) , ` satimage ` , and ` soybean ` . for all other datasets , we reported the average result over 10 random splits , with @xmath176 of the dataset used for training and @xmath177 for testing .
( the splits were the same for all methods . )
error rates ( in % ) of ` tree ` versus ` filter - tree ` ( top ) and all - pairs versus all - pairs filter tree ( top ) on several different datasets with a decision tree or logistic regression classifier.,title="fig : " ] error rates ( in % ) of ` tree ` versus ` filter - tree ` ( top ) and all - pairs versus all - pairs filter tree ( top ) on several different datasets with a decision tree or logistic regression classifier.,title="fig : " ] if computation is constrained and we can afford only @xmath1 computation per multiclass prediction , the filter tree dominates the tree reduction , as shown in figure [ fig : comparison ]
. if computation is relatively unconstrained , all - pairs and the all - pairs filter tree are reasonable choices .
the comparison in figure [ fig : comparison ] shows that there the all - pairs filter tree yields similar prediction performance while using only @xmath10 computation instead of @xmath12 .
test error rates using decision trees ( j48 ) and logistic regression as binary classifier learners are reported in table [ t1 ] , using weka s implementation with default parameters @xcite .
the lowest error rate in each row is shown in bold , although in some cases the difference is insignificant .
in this section , we extend filter trees to @xmath19-elimination tournaments , also called @xmath178-error - correcting tournaments . as this section builds on sections [ s : algorithm ] and [ s : analysis ] , understanding them
is required before reading this section . for simplicity , we work with only the multiclass case .
an extension for cost - sensitive multiclass problems is possible using the importance weighting techniques of the previous section . an _ @xmath19-elimination tournament _ operates in two phases .
the first phase consists of @xmath19 single - elimination tournaments over the @xmath0 labels where a label is paired against another label at most once per round .
consequently , only one of these single elimination tournaments has a simple binary tree structure ; see , for example , figure [ fig : me_phase1 ] for an @xmath25 elimination tournament on @xmath24 labels .
there is substantial freedom in how the pairings of the first phase are done ; our bounds depend on the depth of any mechanism which pairs labels in @xmath19 distinct single elimination tournaments .
one such explicit mechanism is given in @xcite .
note that once an example has lost @xmath19 times , it is eliminated and no longer influences training at the nodes closer to the root . the second phase is a final elimination phase , where we select the winner from the @xmath19 winners of the first phase .
it consists of a redundant single - elimination tournament , where the degree of redundancy increases as the root is approached . to quantify the redundancy ,
let every subtree @xmath179 have a _
@xmath180 equal to the number of leaves under the subtree .
first phase winners at the leaves of the final elimination tournament have charge @xmath55 . for any non - leaf node comparing the outputs of subtrees @xmath181 and @xmath182 , the importance weight of a binary example created at the node is set to either @xmath183 or @xmath184 , depending on whether the label comes from @xmath182 or @xmath181 . in tournament applications ,
an importance weight can be expressed by playing games repeatedly where the winner of @xmath181 must beat the winner of @xmath182 @xmath184 times to advance , and vice versa .
when the two labels compared are the same , the importance weight is set to @xmath112 , indicating there is no preference in the pairing amongst the two choices .
an example of a @xmath26-elimination tournament on @xmath24 players .
there are @xmath25 distinct single elimination tournaments in first phase one in black , one in blue , and one in red .
after that , a final elimination phase occurs over the three winners of the first phase .
the final elimination tournament has an extra weighting on the nodes , detailed in the text . ]
a key concept throughout this section is the _ importance depth _ , defined as the worst - case length ( number of games ) of the overall tournament , where importance - weighted matches in the final elimination phase are played as repeated games . in theorem
[ thm : importance - depth ] we prove a bound on the importance depth .
the computational bound per example is essentially just the importance depth .
[ thm : structural - depth ] _
( structural depth bound ) _ for any @xmath19-elimination tournament , the training and test computation is @xmath185 per example .
the proof is by simplification of the importance depth bound ( theorem [ thm : importance - depth ] ) , which bounds the sum of importance weights at all nodes in the tournament . to see that the importance depth controls the computation , first note that the importance depth bounds the tournament depth since all importance weights are at least 1 . at training time , any one example is used at most once per tournament level starting at the leaves . at testing time , an unlabeled example can have its label determined by traversing the structure from root to leaf .
our regret theorem is the analogue of corollary [ cor : multi ] for error - correcting tournaments , and the notation is as defined there . as in the previous section
, the reduction transforms a multiclass distribution @xmath33 into an induced distribution @xmath101 over binary labeled examples . as before
, @xmath51 denotes the multiclass classifier induced by a given binary classifier @xmath39 and tournament structure @xmath66 .
it is useful to have the notation @xmath186 for the smallest power of @xmath56 larger than or equal to @xmath19 .
[ multi - theorem ] _ ( main theorem ) _ for all distributions @xmath33 over @xmath0-class examples , all binary classifiers @xmath39 , all @xmath19-elimination tournaments @xmath66 , the ratio of @xmath187 to @xmath188 is upper bounded by @xmath189 the first case shows that a regret ratio of @xmath26 is achievable for very large @xmath19 .
the second case is the best bound for cases of common interest .
for @xmath29 it gives a ratio of @xmath30 .
the proof holds for each input @xmath45 , and hence in expectation over @xmath45 .
fix @xmath45 , and let @xmath190 for @xmath191 .
we can define the regret of any label @xmath50 as @xmath192 , where @xmath193 .
the regret of a node @xmath68 comparing labels @xmath78 and @xmath79 from subtrees @xmath181 and @xmath182 , and outputting @xmath78 , is @xmath194 where we use the predicate @xmath195 .
thus @xmath121 is 0 if @xmath68 outputs the more likely label . if @xmath68 is in a first phase tournament , @xmath196 .
finally , the regret of a subtree @xmath66 is defined as @xmath197 . the first part of the proof is by induction on the tree structure @xmath198 of the final phase .
the invariant for a subtree @xmath179 of @xmath198 won by label @xmath78 is @xmath199 where @xmath200 is the winner of a first phase single - elimination tournament @xmath201 . when @xmath179 is a leaf @xmath200 of @xmath198 , we have @xmath202 , where the inequality is from corollary [ cor : multi ] noting that the depth of @xmath201 times the average regret over the nodes in @xmath201 is @xmath203 .
assume inductively that the hypothesis holds at node @xmath68 comparing labels @xmath78 and @xmath79 from subtrees @xmath181 and @xmath182 , and outputting @xmath78 : @xmath204 and @xmath205 we have @xmath206 by the inductive hypothesis .
now , there are two cases : either @xmath207 , in which case @xmath208 and @xmath209 , as desired
. or @xmath210 , in which case @xmath211 and thus @xmath212 finishing the induction . finally , letting @xmath50 be the prediction of @xmath51 on @xmath45 , @xmath213 where @xmath97 is the maximum importance depth . applying the importance depth theorem ( theorem
[ thm : importance - depth ] ) and algebra completes the proof .
the depth bound follows from the following three lemmas .
[ lem : fpdb ] _
( first phase depth bound ) _ the importance depth of the first phase tournament is bounded by the minimum of @xmath214 the depth of the first phase is bounded by the classical problem of robust minimum finding with low depth .
the first three cases hold because any such construction upper bounds the depth of an error - correcting tournament , and one such construction has these bounds @xcite . for the fourth case
, we construct the depth bound by analyzing a continuous relaxation of the problem .
the relaxation allows the number of labels remaining in each single elimination tournament of the first phase to be broken into fractions .
relative to this version , the actual problem has two important discretizations : 1 .
when a single - elimination tournament has only a single label remaining , it enters the next single elimination tournament .
this can have the effect of _ decreasing _ the depth compared to the continuous relaxation .
2 . when a single - elimination tournament has an odd number of labels remaining , the odd label does not play that round .
thus the number of players does not quite halve , potentially _ increasing _ the depth compared to the continuous relaxation . in the continuous version , tournament @xmath6 on round
@xmath97 has @xmath215 labels , where the first tournament corresponds to @xmath216 .
consequently , the number of labels remaining in any of the tournaments is @xmath217 we can get an estimate of the depth by finding the value of @xmath97 such that this number is 1 . this value of @xmath97 can be found using the chernoff bound .
the probability that a coin with bias @xmath218 has @xmath21 or fewer heads in @xmath97 coin flips is bounded by @xmath219 , and the probability that this occurs in @xmath0 attempts is bounded by @xmath0 times that . setting this value to @xmath55
, we get @xmath220 solving the equation for @xmath97 , gives @xmath221 .
this last formula was verified computationally for @xmath222 and @xmath223 by discretizing @xmath0 into factors of @xmath56 and running a simple program to keep track of the number of labels in each tournament at each level . for @xmath224
, we used a pessimistic value of @xmath225 in the above formula to compute the bound , and compared it to the output of the program for @xmath226 .
[ lem : spdb ] _
( second phase depth bound ) _ in any @xmath19-elimination tournament , the second phase has importance depth at most @xmath227 rounds for @xmath228 .
when two labels are compared in round @xmath229 , the importance weight of their comparison is at most @xmath230 .
thus we have @xmath231 .
putting everything together gives the importance depth theorem .
[ thm : importance - depth ] _
( importance depth bound ) _ for all @xmath19-elimination tournaments , the importance depth is upper bounded by @xmath232 we simply add the depths of the first and second phases from lemmas [ lem : fpdb ] and [ lem : spdb ] . for the last case , we bound @xmath233 and eliminate subtractions in lemma [ lem : spdb ] .
all of our lower bounds hold for a somewhat more powerful adversary which is more natural in a game playing tournament setting .
in particular , we disallow reductions which use importance weighting on examples , or equivalently , all importance weights are set to @xmath55 .
note that we can modify our upper bound to obey this constraint by transforming final elimination comparisons with importance weight @xmath6 into @xmath234 repeated comparisons and use the majority vote .
this modified construction has an importance depth which is at most @xmath19 larger implying the ratio of the adversary and the reduction s regret increases by at most @xmath55 .
the first lower bound says that for any reduction algorithm @xmath182 , there exists an adversary @xmath181 with the average per - round regret @xmath9 such that @xmath181 can make @xmath182 incur regret @xmath235 even if @xmath182 knows @xmath9 in advance .
thus an adversary who corrupts half of all outcomes can force a maximally bad outcome . in the bounds below
, @xmath236 denotes the multiclass classifier induced by a reduction @xmath182 using a binary classifier @xmath39 . for any deterministic reduction @xmath182 from @xmath237 classification to binary classification
, there exists a choice of @xmath33 and @xmath39 such that @xmath238 .
the adversary @xmath181 picks any two labels @xmath6 and @xmath239 .
all comparisons involving @xmath6 but not @xmath239 , are decided in favor of @xmath6 .
similarly for @xmath239 .
the outcome of comparing @xmath6 and @xmath239 is determined by the parity of the number of comparisons between @xmath6 and @xmath239 in some fixed serialization of the algorithm .
if the parity is odd , @xmath6 wins ; otherwise , @xmath239 wins .
the outcomes of all other comparisons are picked arbitrarily .
suppose that the algorithm halts after some number of queries @xmath240 between @xmath6 and @xmath239 .
if neither @xmath6 nor @xmath239 wins , the adversary can simply assign probability @xmath218 to @xmath6 and @xmath239 .
the adversary pays nothing while the algorithm suffers loss 1 , yielding a regret ratio of @xmath241 .
assume without loss of generality that @xmath6 wins .
the depth of the tournament is either @xmath240 or at least @xmath242 , because each label can appear at most once in any round .
if the depth is @xmath240 , then since @xmath237 , some label is not involved in any query , and the adversary can set the probability of that label to @xmath55 resulting in @xmath243 . otherwise , @xmath181 can set the probability of label @xmath239 to be @xmath55 while all others have probability @xmath112 .
the total regret of @xmath181 is at most @xmath244 , while the regret of the winning label is @xmath55 .
multiplying by the depth bound @xmath242 , gives a regret ratio of at least @xmath56 .
note that the number of rounds in the above bound can depend on @xmath181 .
next , we show that for any algorithm @xmath182 taking the same number of rounds for any adversary , there exists an adversary @xmath181 with a regret of roughly one third , such that @xmath181 can make @xmath182 incur the maximal loss , even if @xmath182 knows the power of the adversary . for any deterministic reduction @xmath182 to binary classification with number of rounds independent of the query outcomes
, there exists a choice of @xmath33 and @xmath39 such that @xmath245 .
let @xmath182 take @xmath246 rounds to determine the winner , for any set of query outcomes .
we will design an adversary @xmath181 with incurs regret @xmath247 , such that @xmath181 can make @xmath182 incur the maximal loss of 1 , even if @xmath182 knows @xmath9 . the adversary s query answering strategy is to answer consistently with label @xmath55 winning for the first @xmath248 rounds , breaking ties arbitrarily .
the total number of queries that @xmath182 can ask during this stage is at most @xmath249 since each label can play at most once in every round , and each query occupies two labels .
thus the total amount of regret at this point is at most @xmath249 , and there must exist a label @xmath6 other than label @xmath0 with at most @xmath9 losses . in the remaining @xmath250 rounds
, @xmath181 answers consistently with label @xmath6 and all other skills being 0 .
now if @xmath182 selects label @xmath55 , @xmath181 can set @xmath251 with @xmath252 average regret from the first stage .
if @xmath182 selects label @xmath6 instead , @xmath181 can choose that @xmath253 . since the number of queries between labels @xmath6 and @xmath0 in the second stage is at most @xmath9 , the adversary can incurs average regret at most @xmath252 . if @xmath182 chooses any other label to be the winner , the regret ratio is unbounded .
00 m. adler , p. gemmell , m. harchol - balter , r. karp , and c. kenyon .
selection in the presence of noise : the design of playoff systems , soda 1994 .
e. allwein , r. schapire , and y. singer .
reducing multiclass to binary : a unifying approach for margin classifiers , _ journal of machine learning research _ , 1 : 113141 , 2000 . j. aslam and a. dhagat . searching in the presence of linearly bounded errors , stoc 1991 . c. blake and c. merz , uci repository of machine learning databases , university of california , irvine .
r. borgstrom , s. rao kosaraju .
comparison - base search in the presence of errors , stoc 1993 .
p. denejko , k. diks , a. pelc , and m. piotrow .
reliable minimum finding comparator networks , _ fundamenta informaticae _ , 42 : 235249 , 2000 .
t. dietterich and g. bakiri .
solving multiclass learning problems via error - correcting output codes , _ journal of artificial intelligence research _ , 2 : 263286 , 1995 .
u. feige , d. peleg , p. raghavan , and e. upfal .
computing with unreliable information , _ symposium on theory of computing _
, 128137 , 1990 .
d. foster and d. hsu , ` http://hunch.net/?p=468 ` .
_ applied regression analysis , linear models , and related methods _ , sage publications , 1997 .
v. guruswami and a. sahai .
multiclass learning , boosting , and error correcting codes , colt 1999 .
t. hastie and r. tibshirani .
classification by pairwise coupling , nips 1997 .
r. herbrich , t. minka , and t. graepel .
trueskill(tm ) : a bayesian skill rating system , nips 2007 .
d. hsu , j. langford , s. kakade , and t. zhang , multi - label prediction via compressed sensing , arxiv:0902.1284v1 , 2009 .
j. langford and a. beygelzimer .
sensitive error correcting output codes , colt 2005 .
j. langford and b. zadrozny , estimating class membership probabilities using classifier learners , aistat 2005 .
b. ravikumar , k. ganesan , and k. b. lakshmanan . on selecting the largest element in spite of erroneous information , lecture notes in computer science , 247 : 8899 , 1987 .
b. williamson , personal communication .
i. witten and e. frank .
data mining : practical machine learning tools with java implementations , 2000 : ` http://www.cs.waikato.ac.nz/ml/weka/ ` .
a. c. yao and f. f. yao . on fault - tolerant networks for sorting .
_ siam journal of computing _ , 14(1 ) : 120128 , 1985 .
b. zadrozny , j. langford , and n. abe .
cost - sensitive learning by cost - proportionate example weighting , icdm 2003 .
[ cols= " < , > , > , > , > , > , > , > , > , > , > " , ] | text of abstract we present a family of pairwise tournaments reducing @xmath0-class classification to binary classification . these reductions are provably robust against a constant fraction of binary errors , simultaneously matching the best possible computation @xmath1 and regret @xmath2 .
the construction also works for robustly selecting the best of @xmath0-choices by tournament .
we strengthen previous results by defeating a more powerful adversary than previously addressed while providing a new form of analysis . in this
setting , the error correcting tournament has depth @xmath1 while using @xmath3 comparators , both optimal up to a small constant
. reductions , multiclass classification , cost - sensitive learning , tournaments , robust search |
this paper presents motion planning algorithms for kinematic models of underactuated mechanical systems ; we consider kinematic ( i.e. , driftless ) models that are switched control systems , that is , dynamical systems described by a family of admissible vector fields and a control strategy that governs the switching between them . in particular , we focus on families of left - invariant vector fields defined on rigid displacements subgroups .
this class of systems arises in the context of kinematic modeling and kinematic reductions for mechanical control systems ; see the recent works @xcite .
clearly , the transcription into kinematic models simplifies the motion planning problem ; e.g. , @xcite discusses 3r planar manipulators and @xcite discuss the snakeboard system .
motion planning for kinematic models , sometimes referred to as driftless or nonholonomic models , is a classic problem in robotics ; see @xcite and also the references therein . in particular , the algorithms in @xcite focus on dynamical aspects and exploit controllability properties .
for the class of systems of interest in this paper , the search for a motion planning algorithm is closely related to the inverse - kinematics problem .
example inverse - kinematics methods include ( i ) iterative numerical methods for nonlinear optimization , see @xcite , ( ii ) geometric and decoupling methods for classes of manipulators , see @xcite , ( iii ) the paden - kahan subproblems approach , see @xcite , and ( iv ) the general polynomial programming approach , see @xcite .
the latter and more general method is based on tools from algebraic geometry and relies on simultaneously solving systems of algebraic equations . despite these efforts ,
no general methodology is currently available to solve these problems in closed - form .
accordingly , it is common to provide and catalog closed - form solutions for classes of relevant example systems ; see @xcite .
we consider left - invariant control systems evolving on a matrix lie subgroup @xmath3 .
examples include systems on @xmath0 , @xmath1 and @xmath4 . as usual in lie group theory
, we identify left - invariant vector fields with their value at the identity . given a family of left - invariant vector fields @xmath5 on @xmath6 , consider the associated driftless control system @xmath7 where @xmath8 and where the controls @xmath9 take value in @xmath10 . for these systems ,
controllability can be assessed by algebraic means : it suffices to check the lack of involutivity of @xmath11 . recall that for matrix lie algebras , lie brackets are matrix commutators @xmath12 = ab - ba$ ] .
this paper addresses the problem of how to compute feasible motion plans for the control system by concatenating a finite number of flows along the input vector fields .
we call a flow along any input vector field a _ motion primitive _ and its duration a _ coasting time_. therefore , motion planning is reduced to the problem of selecting a finite - length combination of @xmath13 motion primitives @xmath14 and computing appropriate coasting times @xmath15 that steer the system from the identity in the group to any target configuration @xmath16 . in mathematical terms , we need to solve @xmath17 hence , motion planning is transcribed into low - dimensional combinatorial selection and inverse - kinematics problems .
the contribution of this paper is a catalog of solutions for underactuated example systems defined on @xmath0 , @xmath1 , or @xmath2 .
based on a controllability analysis , we classify families of underactuated systems that pose qualitatively different planning problems . for each family
, we solve the planning problem by providing _ a _ combination of @xmath13 motion primitives and corresponding closed - form expressions for the coasting times .
in each case , we attempt to select @xmath18 : generically , this is the minimum necessary ( but sometimes not sufficient ) number of motion primitives needed . if the motion planning algorithm entails exactly @xmath19 motion primitives , i.e. , minimizes the number of switches , we will refer to it as a _ switch - optimal _ algorithm .
sections [ sec : se2 ] , [ sec : so3 ] , and [ sec : se2timesr ] present switch - optimal planners for @xmath0 , @xmath1 , and @xmath2 , respectively . here
we briefly collect the notation used throughout the paper .
let @xmath20 be a set , @xmath21 denote the identity map on @xmath20 and let @xmath22 denote the characteristic function of @xmath20 , i.e. , @xmath23 if @xmath24 and @xmath25 if @xmath26 .
let @xmath27 denote the arctangent of @xmath28 taking into account which quadrant the point @xmath29 is in .
we make the convention @xmath30 .
let @xmath31 be the sign function , i.e. , @xmath32 if @xmath33 , @xmath34 if @xmath35 and @xmath36 .
let @xmath37 be the @xmath38 element of the matrix @xmath39 .
given @xmath40 , let @xmath41 $ ] denote the angle between them .
let @xmath42 denote the euclidean norm
. given a family of left - invariant vector fields @xmath5 on @xmath6 , we associate to each multiindex @xmath43 the forward - kinematics map @xmath44 given by @xmath45 .
let @xmath46 be the basis of @xmath47 : @xmath48 then , @xmath49=e_y$ ] , @xmath50 = e_x$ ] and @xmath51=0 $ ] . for ease of presentation , we write @xmath52 as @xmath53 , and @xmath54 as @xmath55 with this notation , @xmath56 is @xmath57 for @xmath58 , and @xmath59 .
( controllability conditions ) .
[ lem : ctrl - se2 ] consider two left - invariant vector fields @xmath60 and @xmath61 in @xmath47 .
their lie closure is full rank if and only if @xmath62 or @xmath63 .
given @xmath64 = ( 0,c_1 a_2 - a_1 c_2 \,,\ , a_1 b_2 - b_1 a_2)$ ] , one can see that @xmath65\right\ } } = { { { \ensuremath{\mathfrak{se}}}}(2)}$ ] if and only if @xmath66 let @xmath60 and @xmath67 satisfy the controllability condition in lemma [ lem : ctrl - se2 ] .
accordingly , either @xmath68 or @xmath69 is different from zero . without loss of generality
, we will assume that @xmath70 , and take @xmath71 .
as a consequence of lemma [ lem : ctrl - se2 ] , there are two qualitatively different cases to be considered : @xmath72 since @xmath73 , we need at least three motion primitives along the flows of @xmath74 to plan any motion between two desired configurations .
consider the map @xmath75 . in the following propositions , we compute solutions for @xmath76 and @xmath77-systems .
( inversion for @xmath76-systems on @xmath0 ) .
[ prop : se(2)-case - a ] let @xmath78 .
consider the map @xmath79\colon{{\ensuremath{\mathrm{se}(2)}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^3}$ ] , @xmath80 ( \theta , x , y ) = ( { \operatorname{arctan2}\left(\alpha,\beta\right ) } , \rho , \theta -{\operatorname{arctan2}\left(\alpha,\beta\right)}),\ ] ] where @xmath81 and @xmath82 then , @xmath83 $ ] is a global right inverse of @xmath84 , that is , it satisfies @xmath85 = { \operatorname{id}}_{{{\ensuremath{\mathrm{se}(2)}}}}\colon{{\ensuremath{\mathrm{se}(2)}}}\rightarrow{{\ensuremath{\mathrm{se}(2)}}}}$ ] .
note that the algorithm provided in the proposition is not only switch - optimal , but also works globally .
the proof follows from the expression of the map @xmath84 .
let @xmath86 , @xmath87 the equation in @xmath88^t$ ] can be rewritten as @xmath89^t = [ \cos t_1 , \sin t_1]^t t_2 $ ] .
the selection @xmath90 , @xmath91 solves this equation .
( inversion for @xmath77-systems on @xmath0 ) .
[ prop : se(2)-case - b ] let @xmath92 .
define the neighborhood of the identity in @xmath0 @xmath93 consider the map @xmath94\colonu\subset{{\ensuremath{\mathrm{se}(2)}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^3}$ ] whose components are @xmath95_1 ( \theta , x , y ) & = { \operatorname{arctan2}\left(\rho,\sqrt{4 -\rho^2}\right ) } + { \operatorname{arctan2}\left(\alpha,\beta\right ) } \ , , \\ { { \mathcal{ik}}}[{\mathcal{s}}_2]_2 ( \theta , x , y ) & = { \operatorname{arctan2}\left(2-\rho^2,\rho \sqrt{4-\rho^2}\right ) } \ ! , \\ { { \mathcal{ik}}}[{\mathcal{s}}_2]_3 ( \theta , x , y ) & = \theta -{{\mathcal{ik}}}[{\mathcal{s}}_2]_1 ( \theta , x , y)-{{\mathcal{ik}}}[{\mathcal{s}}_2]_2 ( \theta , x , y ) \ , , \end{aligned}\ ] ] and @xmath81 and @xmath96 then , @xmath97 $ ] is a local right inverse of @xmath84 , that is , it satisfies @xmath98 = { \operatorname{id}}_{u}\colonu\rightarrowu}$ ] . if @xmath99 , then @xmath100 and hence @xmath97 $ ] is well - defined on @xmath101 .
let @xmath97 ( \theta , x , y)=(t_1,t_2,t_3)$ ] .
the components of @xmath102 are @xmath103 in an analogous way to the previous proof , one verifies @xmath104 .
the map @xmath97 $ ] in proposition [ prop : se(2)-case - b ] is a local right inverse to @xmath84 on a domain that strictly contains @xmath101 . in other words ,
our estimate of the domain of @xmath97 $ ] is conservative .
for instance , for points of the form @xmath105 , it suffices to ask for @xmath106 for a point @xmath107 , it suffices to ask for @xmath108 additionally , without loss of generality , it is convenient to assume that the vector fields @xmath109 , @xmath110 satisfy @xmath111 , so as to maximize the domain @xmath101 .
we illustrate the performance of the algorithms in fig .
[ fig : se2 ] . and @xmath77-systems
. the parameters of both systems are @xmath112 , @xmath113 .
the target location is @xmath114 .
initial and target location are depicted in dark gray.,title="fig : " ] and @xmath77-systems .
the parameters of both systems are @xmath112 , @xmath113 .
the target location is @xmath114 .
initial and target location are depicted in dark gray.,title="fig : " ]
let @xmath115 be the basis of @xmath116 : @xmath117 here we make use of the notation @xmath118 based on the lie algebra isomorphism @xmath119)$ ] .
rodrigues formula @xcite for the exponential @xmath120 is @xmath121 the commutator relations are @xmath122 = -\widehat{e}_y$ ] , @xmath123 = \widehat{e}_x$ ] and @xmath124 = \widehat{e}_z$ ] .
( controllability conditions ) .
[ lem : ctrl - so3 ] consider two left - invariant vector fields @xmath125 and @xmath126 in @xmath116 .
their lie closure is full rank if and only if @xmath127 or @xmath128 or @xmath129 . given the equality @xmath130 = \widehat{v_1 \times v_2}$ ] , with @xmath131 , one can see that @xmath65\right\ } } = { { { \ensuremath{\mathfrak{so}}}}(3)}$ ] if and only if @xmath132 let @xmath109 , @xmath110 satisfy the controllability condition in lemma [ lem : ctrl - so3 ] . without loss of generality
, we can assume @xmath133 ( otherwise we perform a suitable change of coordinates ) , and @xmath134 .
in what follows , we let @xmath135 . since @xmath136 and @xmath110 are linearly independent ,
necessarily @xmath137 and @xmath138 . since @xmath139 , we need at least three motion primitives to plan any motion between two desired configurations .
consider the map @xmath140 , that is @xmath141 observe that equation is similar to the formula for certain sets of euler angles ; see @xcite .
( inversion for systems on @xmath1 ) .
let @xmath142 and @xmath143 , with @xmath144 and @xmath145 .
define the neighborhood of the identity in @xmath1 @xmath146\}}. \end{aligned}\ ] ] consider the map @xmath147 whose components are @xmath148 where , for @xmath149 , @xmath150 then , @xmath151 is a local right inverse of @xmath84 , that is , it satisfies @xmath152
. let @xmath153 .
then , @xmath154 , and hence @xmath155 is well - defined . denote @xmath156 and let us show @xmath157 .
recall that the rows ( resp .
the columns ) of a rotation matrix consist of orthonormal vectors in @xmath158 .
therefore , the matrix @xmath159 is determined by its third column @xmath160 and its third row @xmath161 .
the factors in admit the following closed - form expressions . for @xmath162 and @xmath163 , @xmath164 and @xmath165
equals @xmath166 now , using the fact that @xmath167 , we get @xmath168 a similar computation shows that @xmath169 , which concludes the proof . if @xmath170 and @xmath110 are perpendicular , then @xmath171 and the map @xmath151 is a global right inverse of @xmath84 .
otherwise , let us provide an equivalent formulation of the constraint @xmath172 $ ] in terms of the axis / angle representation of the rotation matrix @xmath173 . recall that there always exist a , possibly non - unique , rotation angle @xmath174 $ ] and an unit - length axis of rotation @xmath175 such that @xmath176 . because @xmath177 , an equivalent statement of rodrigues formula is @xmath178 from @xmath179 , we compute @xmath180 therefore , @xmath172 $ ] if and only if @xmath181 two sufficient conditions are also meaningful . in terms of the rotation angle ,
if @xmath182 then @xmath183 , and in turn equation is satisfied . in terms of the axis of rotation
, a sufficient condition for equation is @xmath184 .
we illustrate the performance of the algorithm in fig .
[ fig : so3 ] . .
the system parameters are @xmath185 .
the target final rotation is @xmath186 . to render the sequence of three rotations visible ,
the body is translated along the inertial @xmath187-axis . ]
let @xmath188 be a basis of @xmath189 , where @xmath190 stands for the basis of @xmath47 introduced in section [ sec : se2 ] . with a slight abuse of notation ,
we let @xmath191 denote @xmath192 , and we similarly redefine @xmath193 and @xmath194 . we also let
the only non - vanishing lie algebra commutators are @xmath49 = e_y$ ] and @xmath196 = - e_x$ ] .
a left - invariant vector field @xmath197 in @xmath189 is written as @xmath198 , and @xmath199 as @xmath200 .
the exponential map , @xmath201 , is given component - wise by the exponential on @xmath47 and @xmath202 , respectively .
that is , @xmath203 is equal to @xmath204 if @xmath58 , and @xmath205 if @xmath206 .
[ lem : ctrl - se2timesr - two ] ( controllability conditions for systems in @xmath2 with 2 inputs ) .
consider two left - invariant vector fields @xmath207 and @xmath208 in @xmath189 .
their lie closure is full rank if and only if @xmath209 , and either @xmath63 or @xmath210 . since @xmath64 = ( 0\,,c_1 a_2 - a_1 c_2 \,,\ , a_1 b_2 - b_1 a_2 \ , , 0)\neq 0 $ ] , we deduce that either @xmath63 or @xmath211 .
in particular , this implies that necessarily @xmath70 or @xmath212 .
assume @xmath70 .
now , @xmath213 = ( 0 , a_1(-b_2 a_1 + b_1 a_2 ) , a_1(c_1 a_2 - c_2 a_1),0 ) \ , , \end{aligned}\ ] ] and note that @xmath214 = ( a_2/a_1 ) [ v_1,[v_1,v_2]]$ ] .
finally , @xmath215 if and only if @xmath216 \neq 0 . \end{gathered}\ ] ] since @xmath64 \neq 0 $ ] , this reduces to @xmath209 .
let @xmath109 , @xmath110 satisfy the controllability condition in lemma [ lem : ctrl - se2timesr - two ] .
without loss of generality , we can assume @xmath217 .
as in the case of @xmath0 , there are two qualitatively different situations to be considered : @xmath218 [ lem : ctrl - se2timesr - three ] ( controllability conditions for @xmath2 systems with 3 inputs ) .
consider three left - invariant vector fields @xmath219 , @xmath220 in @xmath189 .
assume @xmath221 , for @xmath222 and @xmath223 .
then , possibly after a reordering of the vector fields , they must fall in one of the following cases : @xmath224 without loss of generality , we can assume that @xmath64 \neq 0 $ ] and @xmath71 . since @xmath225 , then @xmath226 .
given that the lie closure of @xmath227 is full - rank , and @xmath228\right\}})= 3 $ ] , we have that @xmath229 .
this latter fact , together with @xmath230 , implies that @xmath231 = 0 $ ] , and therefore @xmath232 , @xmath233 .
we now distinguish two situations depending on @xmath234 $ ] being zero or not . _
( a ) @xmath234 \neq 0$]_. necessarily , @xmath235 .
therefore , we can assume @xmath236 . since @xmath237 is not full - rank , then @xmath238 .
we then have a @xmath239-system .
_ ( b ) @xmath234 = 0$]_. necessarily , @xmath240 and @xmath241 . depending on the values of @xmath69 and @xmath242 , we consider : 1 . if @xmath243 , then @xmath244 , @xmath245 , @xmath246 .
then , this is a @xmath247-system .
2 . if @xmath238 , and @xmath236 , then @xmath248 , @xmath249 and also @xmath250 .
this is not possible as it would make @xmath251 .
3 . if @xmath252 and @xmath253 , then @xmath254 , and @xmath255 . therefore ,
this is a @xmath256-system .
4 . finally , if @xmath257 and @xmath236 , then @xmath258 , @xmath259 , and @xmath260 , which makes @xmath109 and @xmath110 linearly dependent . let @xmath109 , @xmath110 satisfy the controllability condition in lemma [ lem : ctrl - se2timesr - two ] . since @xmath261 , we need at least four motion primitives to plan any motion between two desired configurations .
consider the map @xmath262 .
[ prop : se(2)timesr - case - a - impossibility ] ( lack of switch - optimal inversion for @xmath263-systems on @xmath264 ) .
let @xmath265 .
then , the map @xmath266 in not invertible at any neighborhood of the origin .
let @xmath267 .
then , @xmath268 consider a configuration with @xmath269 .
then , the equation in @xmath29 is invertible if and only if the map @xmath270 defined by @xmath271 is invertible . but @xmath272 can not be inverted in @xmath273 , @xmath274 .
an identical negative result holds if we start taking motion primitives along the flow of @xmath109 instead of @xmath110 , i.e. , if we consider the map @xmath275 .
consider the map @xmath276 .
( inversion for @xmath263-systems on @xmath264 ) .
[ prop : se(2)timesr - case - a ] let @xmath265 .
consider the map @xmath277\colon{{\ensuremath{\mathrm{se}(2 ) } } } \times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^5}$ ] whose components are @xmath278_1 ( \theta , x , y , z ) & = \pi { { \operatorname{ind}}}_{]-\infty,0 [ } ( \gamma-\rho ) + { \operatorname{arctan2}\left(\alpha,\beta\right ) } \\ & \qquad + { \operatorname{arctan2}\left((\rho+\gamma)/2,0\right ) } , \\ { { \mathcal{ik}}}[{\mathcal{t}}_1]_2 ( \theta , x , y , z ) & = ( \gamma-\rho)/2 , \\ { { \mathcal{ik}}}[{\mathcal{t}}_1]_3 ( \theta , x , y , z ) & = { \operatorname{arctan2}\left((\rho^2-\gamma^2)/4,0\right ) } \\ & + \pi \left ( { { \operatorname{ind}}}_{]-\infty,0 [ } ( \gamma+\rho ) -{{\operatorname{ind}}}_{]-\infty,0 [ } ( \gamma-\rho ) \right ) , \\ { { \mathcal{ik}}}[{\mathcal{t}}_1]_4 ( \theta , x , y , z ) & = ( \gamma+\rho)/2 , \\ { { \mathcal{ik}}}[{\mathcal{t}}_1]_5 ( \theta , x , y , z ) & = \ ! \theta \ ! - \ !
{ { \mathcal{ik}}}[{\mathcal{t}}_1]_1 ( \theta , x , y , z ) \ ! - \ !
{ { \mathcal{ik}}}[{\mathcal{t}}_1]_3 ( \theta , x , y , z ) , \end{aligned}\ ] ] where
@xmath81 and @xmath279 then , @xmath280 $ ] is a global right inverse of @xmath281 , that is , it satisfies @xmath282 = { \operatorname{id}}_{{{\ensuremath{\mathrm{se}(2)}}}\times{{\ensuremath{\mathbb{r}}}}}\colon{{\ensuremath{\mathrm{se}(2)}}}\times{{\ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathrm{se}(2)}}}\times{{\ensuremath{\mathbb{r}}}}}$ ] .
the proof follows from the expression of @xmath281 .
let @xmath283 , @xmath284 the equation in @xmath88^t$ ] can be rewritten as @xmath285 which is solved by the selection of coasting times given by the components of the map @xmath280 $ ] .
( inversion for @xmath286-systems on @xmath264 ) .
[ prop : se(2)timesr - case - b ] let @xmath287 .
define the neighborhood of the identity in @xmath4 @xmath288 consider the map @xmath289\colon{{\ensuremath{\mathrm{se}(2 ) } } } \times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^5}$ ] whose components are @xmath290_1 ( \theta , x , y , z ) & = { \operatorname{arctan2}\left(l,\sqrt{4 - l^2}\right ) } + { \operatorname{arctan2}\left(\alpha,\beta\right ) } , \\
{ { \mathcal{ik}}}[{\mathcal{t}}_2]_2 ( \theta , x , y , z ) & = 2 { \operatorname{arctan2}\left(\sqrt{4-l^2},l\right ) } , \\ { { \mathcal{ik}}}[{\mathcal{t}}_2]_3 ( \theta , x , y , z ) & = -{\operatorname{arctan2}\left(\rho - l,\sqrt{4-(\rho - l)^2}\right ) } \\ & - { { \mathcal{ik}}}[{\mathcal{t}}_2]_1 ( \theta , x , y , z ) - { { \mathcal{ik}}}[{\mathcal{t}}_2]_2 ( \theta , x , y , z ) , \\ { { \mathcal{ik}}}[{\mathcal{t}}_2]_4 ( \theta , x , y , z ) & = \gamma - { { \mathcal{ik}}}[{\mathcal{t}}_2]_2 ( \theta , x , y , z ) , \\ { { \mathcal{ik}}}[{\mathcal{t}}_2]_5 ( \theta , x , y , z ) & = \theta - \sum_{i=1}^4 { { \mathcal{ik}}}[{\mathcal{t}}_2]_i ( \theta , x , y , z ) , \end{aligned}\ ] ] where
@xmath81 , @xmath291 , @xmath292 and @xmath293 then , @xmath294 $ ] is a local right inverse of @xmath281 , that is , it satisfies @xmath295 = { \operatorname{id}}_{u}\colonu\rightarrowu}$ ] . if @xmath296 , then @xmath297 and @xmath298
this in turn implies that @xmath299 over @xmath297 .
the second inequality guarantees that @xmath300 is well - defined .
the first one implies @xmath301 $ ] , which makes @xmath294 $ ] well - defined on @xmath101 .
let @xmath294 ( \theta , x , y , z)=(t_1,t_2,t_3,t_4,t_5)$ ] . the components of @xmath302 are the following @xmath303 after some rather involved computations , one can verify @xmath304 .
let @xmath109 , @xmath110 , @xmath305 satisfy the controllability condition in lemma [ lem : ctrl - se2timesr - three ] .
consider @xmath306 .
( inversion for @xmath307-systems on @xmath264 ) .
[ prop : se(2)timesr - case - c ] let @xmath308 .
consider the map @xmath309\colon{{\ensuremath{\mathrm{se}(2 ) } } } \times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^4}$ ] whose components are @xmath310_1 ( \theta , x , y , z ) & = { \operatorname{arctan2}\left(\alpha,\beta\right ) } - { { \mathcal{ik}}}[{\mathcal{t}}_3]_2 ( \theta , x , y , z ) \ , , \\ { { \mathcal{ik}}}[{\mathcal{t}}_3]_2 ( \theta , x , y , z ) & = \frac{z - d_1 \theta}{d_3-d_1 } , \\ { { \mathcal{ik}}}[{\mathcal{t}}_3]_3 ( \theta , x , y , z ) & = \rho \ , , \\ { { \mathcal{ik}}}[{\mathcal{t}}_3]_4 ( \theta , x , y , z ) & = \theta - { \operatorname{arctan2}\left(\alpha,\beta\right)}\ , , \end{aligned}\ ] ] where @xmath81 and @xmath311 then , @xmath312 $ ] is a global right inverse of @xmath313 , that is , it satisfies @xmath314 = { \operatorname{id}}_{{{\ensuremath{\mathrm{se}(2)}}}\times { { \ensuremath{\mathbb{r}}}}}\colon{{\ensuremath{\mathrm{se}(2)}}}\times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathrm{se}(2)}}}\times { { \ensuremath{\mathbb{r}}}}}$ ] .
the proof follows from the expression of @xmath313 . if @xmath315 , then @xmath316 the equation in @xmath88^t$ ] can be rewritten as @xmath317 which is solved by the selection given by @xmath318(\theta , x , y , z)$ ] .
consider the map @xmath319 .
( inversion for @xmath247-systems on @xmath264 ) .
[ prop : se(2)-case - d ] let @xmath320 .
consider the map @xmath321\colon{{\ensuremath{\mathrm{se}(2 ) } } } \times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^4}$ ] given by @xmath322 ( \theta , x , y , z ) = \big ( { \operatorname{arctan2}\left(\alpha,\beta\right ) } , \rho , \\ \theta - { \operatorname{arctan2}\left(\alpha,\beta\right ) } , \frac{z - d_1 \theta}{d_3 } \big ) , \end{gathered}\ ] ] where @xmath81 and @xmath323 then , @xmath324 $ ] is a global right inverse of @xmath325 , that is , it satisfies @xmath326 = { \operatorname{id}}_{{{\ensuremath{\mathrm{se}(2)}}}\times { { \ensuremath{\mathbb{r}}}}}\colon{{\ensuremath{\mathrm{se}(2)}}}\times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathrm{se}(2)}}}\times { { \ensuremath{\mathbb{r}}}}}$ ] . if @xmath327 , then @xmath328 the equation in @xmath88^t$ ] can be rewritten as @xmath89^t = [ \cos t_1 , \sin t_1]^t t_2 $ ] .
as in the proof of proposition [ prop : se(2)-case - a ] , the selection @xmath329 , @xmath330 solves it .
( inversion for @xmath331-systems on @xmath264 ) .
[ prop : se(2)timesr - case - e ] let @xmath332 .
define the neighborhood of the identity in @xmath4 @xmath333 consider the map @xmath334\colonu\subset{{\ensuremath{\mathrm{se}(2 ) } } } \times { { \ensuremath{\mathbb{r}}}}\rightarrow{{\ensuremath{\mathbb{r}}}}^4}$ ] whose components are @xmath335_1 ( \theta , x , y , z ) & = { \operatorname{arctan2}\left(\rho,\sqrt{4 -\rho^2}\right ) } + { \operatorname{arctan2}\left(\alpha,\beta\right ) } \ , , \\ { { \mathcal{ik}}}[{\mathcal{t}}_5]_2 ( \theta , x , y , z ) & = { \operatorname{arctan2}\left(2-\rho^2,\rho \sqrt{4-\rho^2}\right ) } \ ! , \\ { { \mathcal{ik}}}[{\mathcal{t}}_5]_3 ( \theta , x , y , z ) & = \theta -{{\mathcal{ik}}}[{\mathcal{t}}_5]_1 ( \theta , x , y)-{{\mathcal{ik}}}[{\mathcal{t}}_5]_2 ( \theta , x , y ) \ , , \\
{ { \mathcal{ik}}}[{\mathcal{t}}_5]_4 ( \theta , x , y , z ) & = \frac{z - d_1 \theta}{d_3 } \ , , \end{aligned}\ ] ] and @xmath81 and @xmath336 then , @xmath337 $ ] is a local right inverse of @xmath325 , that is , it satisfies @xmath338 = { \operatorname{id}}_{u}\colonu\rightarrowu}$ ] . if @xmath296 , then one can see that @xmath339 , and therefore @xmath337 $ ] is well - defined on @xmath101 .
let @xmath337 ( \theta , x , y , z)=(t_1,t_2,t_3,t_4)$ ] . the components of @xmath340 are @xmath341 one can verify that @xmath342 . we illustrate the performance of the algorithms in fig .
[ fig : se2real ] .
-system on @xmath2 .
the system parameters are @xmath343 , @xmath344 , @xmath345 , @xmath346 , and @xmath347 .
the target location is @xmath348 . ]
we have presented a catalog of feasible motion planning algorithms for underactuated controllable systems on @xmath0 , @xmath1 and @xmath2 .
future directions of research include ( i ) considering other relevant classes of underactuated systems on @xmath349 , ( ii ) computing catalogs of optimal sequences of motion primitives , and ( iii ) developing hybrid feedback schemes that rely on the proposed open - loop planners to achieve point stabilization and trajectory tracking .
k. m. lynch , n. shiroma , h. arai , and k. tanie , `` collision - free trajectory planning for a 3-dof robot with a passive joint , '' _ international journal of robotics research _ , vol .
19 , no . 12 , pp . 11711184 , 2000 .
f. bullo and k. m. lynch , `` kinematic controllability for decoupled trajectory planning in underactuated mechanical systems , '' _ ieee transactions on robotics and automation _
17 , no . 4 , pp .
402412 , 2001 .
f. bullo , a. d. lewis , and k. m. lynch , `` controllable kinematic reductions for mechanical systems : concepts , computational tools , and examples , '' in _ mathematical theory of networks and systems _ , ( notre dame , in ) , aug .
2002 .
d. halperin , l. e. kavraki , and j .- c .
latombe , `` robotics , '' in _ handbook of discrete and computational geometry _
( j. e. goodman and j. orourke , eds . ) , ch .
41 , pp .
755778 , boca raton , fl : crc press , 1997 .
g. lafferriere and h. j. sussmann , `` a differential geometric approach to motion planning , '' in _ nonholonomic motion planning _ ( z. li and j. f. canny , eds . ) , pp .
235270 , boston , ma : kluwer academic publishers , 1993 . | this paper presents motion planning algorithms for underactuated systems evolving on rigid rotation and displacement groups .
motion planning is transcribed into ( low - dimensional ) combinatorial selection and inverse - kinematics problems .
we present a catalog of solutions for all underactuated systems on @xmath0 , @xmath1 and @xmath2 classified according to their controllability properties . |
density - functional theory ( dft ) constitutes one of the most popular methods in quantum chemistry .
the foundations of dft rest in particular on three contributions : first , the hohenberg kohn ( hk ) theorems established a one - to - one mapping between a set of scalar potentials and a set of ground - state densities as well as a variation principle based on the density @xcite . here , the density is the charge density ( strictly , the negative of the charge density in units of the elementary electron charge ) .
second , the levy lieb constrained - search expression provided a formal but explicit expression for the intrinsic energy ( the universal density functional ) and clarified significant fundamental points @xcite .
third , lieb further generalized the universal functional to a convex functional represented in terms of a legendre fenchel transform @xcite . from a mathematical point of view
, lieb s formulation is particularly attractive as it allows application of convex analysis to establish several properties of the intrinsic energy functional @xcite .
additionally , lieb s framework has made feasible practical calculations of approximations to the exact intrinsic energy functional and adiabatic connection curves @xcite , enabling detailed comparisons of the properties of approximate and near - exact density functionals to be made @xcite .
standard dft , involving universal energy functionals of only the charge density , is limited to the treatment of physical systems that may be represented as eigenstates of hamiltonians that differ only in their scalar potentials . to treat systems subject to an external magnetic field
, it is necessary to introduce an additional dependence on the magnetic field or its associated vector potential into the hamiltonian .
consequently , a dependence on a corresponding variable apart from charge density is needed in the universal energy functional . in magnetic - field
density functional theory ( b - dft ) this is resolved by constructing a family of density functionals
one for each external magnetic field@xcite . in the present work
, we consider the alternative current density - functional theory ( cdft ) , where the additional variable is either the paramagnetic current density or the physical current density .
we restrict our attention to non - relativistic formulations and most of the discussion will for simplicity not be concerned with densities or density - contributions arising from spin - degrees of freedom .
we term the variables on which the energy functionals explicitly depend the _ basic variables _ and make a distinction between _ basic densities _ and _ basic potentials_. many choices of basic densities are conceivable @xcite ; we require only that the choices result in useful density - functional theories .
our perspective thus differs from that in recent works on cdft by pan and sahni @xcite , who restrict the term _ basic variable _ to variables that admit an hk theorem .
although it appears naturally in the generic framework introduced by ayers and fuentealba @xcite , the possibility of choosing basic potentials other than the standard electromagnetic potentials and fields has not previously been explored in detail . by far
the most developed form of cdft is that due to vignale and rasolt @xcite , who use the charge and paramagnetic current densities as basic variables . for these variables , a kohn
sham approach has been formulated @xcite with an associated adiabatic - connection @xcite , virial and scaling relations @xcite analogous to standard kohn sham dft .
in addition , optimized - effective - potential ( oep ) approaches based on this formulation of cdft have been presented to treat non - collinear magnetism @xcite and extensions to time - dependent cdft have been considered @xcite . however , in cdft based on the charge and paramagnetic current densities as basic variables , no hk - type theorem exists and the consequences of this have been extensively discussed in the literature @xcite . in the present work
, we examine this question for cdft in some detail , demonstrating how convex analysis of the underlying universal density functional can be a significant aid in clarifying the relationship between basic variables of cdft and the potentials .
a cdft featuring the gauge - invariant physical current density ( rather than the paramagnetic current density ) as a basic variable is appealing from a physical perspective and is therefore also considered here .
specifically , we examine the formulations due to diener @xcite and pan and sahni @xcite .
we begin in sec .
[ sec : prelim ] by introducing notation related to sets of basic potentials , basic densities , and mappings between them . in sec .
[ sec : cdft ] , we consider cdfts that use the charge and paramagnetic current densities as basic variables in particular , sec .
[ sec : restoreconc ] establishes the concavity of a universal density functional based on these variables and sec .
[ subsec : numf ] outlines the opportunities that this formulation affords for numerical studies of this functional .
next , in sec .
[ secphyscdft ] , the use of the charge and physical current densities as basic variables is considered and two previous formulations @xcite are examined . our concluding remarks are presented in sec .
[ sec : conc ] .
before discussing cdft , we briefly review standard dft , with emphasis on lieb s treatment based on convex conjugation @xcite . the concepts and techniques of convex analysis introduced here
are well suited to the study of dft and will later be used in our discussion of cdft .
some background is also given in the appendix .
we consider a system of @xmath0 electrons with an electronic hamiltonian of the form ( in atomic units ) @xmath1 = \frac{1}{2 } \sum_k p_k^2 + \sum_k v(\mathbf{r}_k ) + w , \label{hamv}\ ] ] where @xmath2 is the canonical momentum operator of electron @xmath3 , @xmath4 is the external potential at position @xmath5 , and @xmath6 is the two - electron coulomb repulsion operator .
the state of the system is described by a density matrix @xmath7 , which is a convex combination of normalized @xmath0-electron pure - state density matrices @xmath8 where the wave functions @xmath9 are antisymmetric in the space and spin coordinates @xmath10 of the @xmath0 electrons . the electron density associated with such density matrices
is given by @xmath11 where the volume element is @xmath12 , i.e. , the integration is over all @xmath0 spin and spatial coordinates except @xmath13 .
the ground - state energy is obtained from the rayleigh
ritz variation principle , @xmath14 = \inf_{\gamma } { \mathrm{tr } ( { \gamma h[v ] } ) } , \ ] ] where the minimization is over all @xmath0-electron density matrices .
an infimum rather than a minimum is taken in eq . since @xmath15 may or may not support an @xmath0-electron ground state .
the set of potentials that support one or more @xmath0-electron ground states ( and for which therefore the infimum is attained ) is denoted by @xmath16 ; the potentials in @xmath16 are sometimes said to be @xmath17-representable .
conversely , a density that is an ensemble ground - state density for some potential @xmath18 is said to be ( ensemble ) @xmath15-representable ; the set of @xmath15-representable densities is denoted by @xmath19 . for convenience , we shall also refer to @xmath17-representable potentials and @xmath15-representable densities as ground - state potentials and densities , respectively . in the constrained - search formalism of dft
, we write the rayleigh
ritz variation principle as an hk variation principle , @xmath20 = \inf_{\rho \in \mathcal{i}_n } \left ( f[\rho ] + ( \rho | v ) \right ) , \label{eqhk}\ ] ] where @xmath21 is the set of @xmath0-representable densities that is , the set of the nonnegative densities @xmath17 with @xmath22 and with a finite von weizscker kinetic energy
. the lieb constrained - search functional @xmath23 is given by @xmath24 = \inf_{\gamma\mapsto \rho } { \mathrm{tr}({\gamma h[0 ] } ) } , \label{fll}\ ] ] where the notation @xmath25 indicates that the minimization is restricted to density matrices @xmath7 that reproduce the density @xmath17 .
if @xmath17 is not @xmath0-representable , no such @xmath7 exists , and @xmath26=+\infty$ ] by definition .
the hk variation principle in eq .
is well defined for all potentials @xmath15 that have a finite pairing with every @xmath27 , @xmath28 an especially attractive formulation of dft is lieb s formulation in terms of legendre fenchel transformations or convex conjugation .
this formulation is not only elegant , but also fits naturally in the well - developed mathematical field of convex analysis , allowing application of deep results of convex analysis to dft , of which we will give some examples .
lieb s formulation of dft begins with the observation that the ground - state energy @xmath29 $ ] is upper semi - continuous and concave in @xmath15 and therefore may be represented by its conjugate function : lieb s universal density functional @xmath26 $ ] .
the ground - state energy and density functionals are then related as @xmath30 & = \inf_{\rho\in x } \left[f[\rho ] + ( \rho|v)\right ] , \label{eqliebvfun } \\
f[\rho ] & = \sup_{v\in x^ * } \left[e[v ] - ( \rho|v)\right ] .
\label{eqliebffun}\end{aligned}\ ] ] [ eqlieb ] here , @xmath31 is a banach space ( a complete normed vector space ) that contains @xmath21 , and @xmath32 is its dual that is , the set of all bounded linear functionals on @xmath31 ( also a banach space ) , thereby ensuring that @xmath33 .
lieb identified @xmath34 and @xmath35 , which contains , among others , all coulomb potentials .
the observations that @xmath29 $ ] and @xmath26 $ ] are upper and lower semicontinuous concave and convex functions , respectively , together with the identification of the banach spaces @xmath31 and @xmath32 are the key elements that place dft within the setting of convex analysis . the duality of @xmath36 and @xmath23 apparent in eq .
means that the same information is contained in either functional but encoded in different ways ; this duality is emphasized by referring to @xmath36 and @xmath23 as the extrinsic and intrinsic energies , respectively , of the electronic system .
the infimum and supremum expressions on the right - hand sides of eqs . and feature linear pairings of densities and potentials and are therefore by construction concave and convex , respectively .
indeed , a necessary and sufficient condition for @xmath29 $ ] ( @xmath26 $ ] ) to be upper ( lower ) semicontinuous and concave ( convex ) is the existence of such expressions @xcite . unlike the lieb density - matrix constrained - search functional ,
lieb constrained - search functional , defined in terms of pure states rather than density matrices , is not convex and hence not identical to the lieb functional . in the present paper ,
we generalize lieb s formulation of dft to cdft .
in particular , we discuss the vignale
rasolt constrained - search functional as a generalization of the lieb functional to systems in the presence of a vector potential .
we shall see that such a generalization is possible after a redefinition of the scalar potential .
however , we here leave aside technical questions such as lower semicontinuity and infimum and supremum domains ( i.e. , banach spaces ) , which is the subject of future work ; for a discussion of such mathematical issues within standard dft , see lieb @xcite and eschrig @xcite .
by contrast , the concavity of @xmath36 and the convexity of @xmath23 are essential properties for our discussion of cdft .
if @xmath29 $ ] happens to be non - concave , it can not be represented by an expression like that in eq .
, even by allowing @xmath23 to be non - convex .
therefore , no universal functional with a linear potential pairing can exist for non - concave energies ( such as those of excited states of the same symmetry as the ground state ) .
a cdft with the paramagnetic current as a basic density was considered in the seminal work of vignale and rasolt @xcite . in their formulation of cdft ,
the basic potentials are the standard electromagnetic potentials @xmath37 and the basic densities are the charge density and paramagnetic current density @xmath38 .
we shall here first review their theory and then discuss an alternative formalism , based on a redefinition of the basic scalar potential .
we consider electrons subject to time - independent external electromagnetic fields @xmath39 and @xmath40 , represented by the scalar potential @xmath4 and the vector potential @xmath41 , respectively . for potentials
@xmath42 , we introduce the equivalence relation @xmath43 which defines equivalence classes of potentials that differ only by a static gauge transformation , thereby representing the same external fields .
we note that a general gauge transformation of @xmath15 and @xmath44 is given by @xmath45 and @xmath46 , for some arbitrary gauge function @xmath47 .
if @xmath44 is to remain static after the transformation , we must require that @xmath48 , where @xmath49 is constant .
it follows that a general time - independent gauge transformation is given by @xmath50 and @xmath51 , where the constant @xmath49 and the function @xmath52 are independent .
therefore , the equivalence relation in eq . holds if and only if there exists a constant @xmath49 and a sufficiently well - behaved gauge function @xmath52 such that @xmath53 and @xmath54 . in the presence of a vector potential , the electronic hamiltonian in eq .
is modified by replacing the canonical momentum operator @xmath55 by the mechanical ( kinetic ) momentum operator @xmath56 , yielding @xmath57 = \frac{1}{2 } \sum_k \pi_k^2 + \sum_k v(\mathbf{r}_k ) + w. \label{ham}\ ] ] we have here omitted the spin - dependent term , @xmath58 , from the hamiltonian . by analogy with eq .
, the rayleigh ritz variation principle in the presence of a vector potential is given by @xmath59 = \inf_{\gamma } { \mathrm{tr } ( { \gamma h[v,\mathbf{a } ] } ) } , \ ] ] where the minimization is over density matrices @xmath7 containing @xmath0 electrons , see eq . .
an infimum rather than a minimum is taken to ensure that the energy is well defined also when @xmath37 does not support a ground state .
we denote the set of all potentials @xmath60 that support a ground state with this hamiltonian by @xmath61 \text{\ has a g.s.}\}\ ] ] and also introduce the related set @xmath62 \text{\ has a g.s.}\}\ ] ] in preparation of a reparameterization of the scalar potential that will be introduced later . in the presence of a vector potential , the ensemble ground - state charge densities @xmath17 are as before given by eq . . regarding the induced currents , we distinguish between the paramagnetic current density and the physical current density
the former is defined as @xmath63 the paramagnetic current density is gauge - dependent and unobservable .
the physical current density is given by @xmath64 and satisfies the relation @xmath65 .
unlike the paramagnetic current , the physical current is gauge - invariant .
finally , the sets of paramagnetic and physical @xmath15-representable ground - state densities are denoted by @xmath66\ } , \\
\mathcal{b}_n & = \{(\rho,\mathbf{j } ) | ( \rho,\mathbf{j } ) \text{\ is g.s .
den.\ of some\ } h[v,\mathbf{a}]\},\end{aligned}\ ] ] where both mixed and pure states are allowed . the hk theorem of standard dft states that the ground - state density @xmath17 determines the scalar potential @xmath15 up to a constant shift .
hence , two potentials that differ by more than a constant shift can not give rise to the same ground - state density .
this fact establishes a mapping from ground - state densities to potentials .
vignale and rasolt established that two different potentials @xmath67 with different ground - state wave functions @xmath68 can not give rise to the same paramagnetic ground - state density @xmath38 . however
, this does not establish an analogue of the hk theorem for cdft @xcite since different potentials @xmath67 can map to the same density in @xmath69 via the same wave function , @xmath70 = \psi[v_2,\mathbf{a}_2]$ ] .
vignale and rasolt s result can be extended to a form that applies also when the ground states are degenerate : _ let @xmath71 be a ground state of @xmath72 $ ] and let @xmath73 be a ground state of @xmath74 $ ] .
if @xmath71 and @xmath73 give rise to the same paramagnetic density , @xmath75 , then @xmath73 is also a ground state of @xmath72 $ ] and @xmath71 is also a ground state of @xmath74 $ ] .
_ the above statement appears to be as close as one can get to a hk - like result for paramagnetic densities @xmath38 .
a cdft formulated in terms of the paramagnetic current density thus can not be based on a formal mapping from ground - state densities to potentials . on the other hand , rigorous formulations of standard dft
do not rely on the hk mapping from densities to potentials .
these rigorous formulations can be extended to the paramagnetic current density as a basic variable .
the absence of an hk - type theorem in cdft is therefore not a serious impediment .
an important property of @xmath29 $ ] in eq .
is its concavity in @xmath15 , which established duality with the universal density functional @xmath26 $ ] .
however , unlike @xmath29 $ ] , the energy functional @xmath76 $ ] in eq .
is not concave .
the non - concavity of @xmath76 $ ] is apparent in , for example , any diamagnetic ground state at vanishing external magnetic field .
such a ground state has a negative definite magnetizability tensor @xmath77 and , when restricted to weak uniform magnetic fields @xmath78 , the energy is a convex function @xmath79 in @xmath80 and therefore in @xmath44 . in more detail , consider a one - electron system confined to the ( two - dimensional ) @xmath81-plane , subject both to a uniform magnetic field along the @xmath82-axis and to a harmonic - oscillator potential .
parameterizing the scalar and vector potentials under consideration as @xmath83 we obtain the following hamiltonian @xmath84 & = \frac{1}{2 } p^2 + \frac{1}{2 } b l_z + v_{\text{ho}}(\mathbf{r};k ) + \frac{1}{2 } a_{\perp}(\mathbf{r};b)^2 \\ & = \frac{1}{2 } p^2 + \frac{1}{2 } b l_z + \frac{1}{8 } \left(4k + b^2\right ) ( x^2+y^2 ) , \end{split}\ ] ] where @xmath85 is a good quantum number . for @xmath86 and some finite interval @xmath87 ,
the ground state has @xmath88 and application of a magnetic field has exactly the same effect as the introduction of a harmonic - oscillator potential .
for these potentials , the ground - state energy is @xmath89 = e[v_{\text{ho } } + \tfrac{1}{2 } a_{\perp}^2 , \mathbf{0 } ] = \sqrt{k + \tfrac{1}{4 } b^2}. \label{eq : example - energy}\ ] ] note that the right - hand side is concave in @xmath90 and convex in @xmath91 .
hence , on the restricted set of potentials spanned by @xmath92 and @xmath87 , the functional @xmath93 $ ] is not only non - concave but convex in its second argument . on a larger domain ,
the functional is neither concave nor convex .
we conclude that @xmath94 $ ] can not be represented by a conjugate functional in the manner of eq . .
however , this does not preclude a constrained - search formulation of cdft , as discussed in the next subsection . rewriting the rayleigh
ritz variation principle in eq . by analogy with the constrained - search approach of standard dft in eq .
, we obtain an hk - type variation principle for a system in the presence of a scalar and vector potential , @xmath95 = \inf_{\rho,\mathbf{j}_{\text{p } } } \left [ f_{\text{vr}}[\rho,\mathbf{j}_{\text{p } } ] + ( \rho|v+\tfrac{1}{2 } a^2 ) + ( \mathbf{j}_{\text{p}}|\mathbf{a } ) \right],\ ] ] where the _ vignale rasolt constrained - search functional _ is given by @xmath96 = \inf_{\gamma\mapsto(\rho,\mathbf{j}_{\text{p } } ) } { \mathrm{tr}({\gamma ( \tfrac{1}{2 } p^2 + w ) } ) } \label{fvr}\ ] ] and we have introduced the following notation for the pairing between a current density and a vector potential : @xmath97 like the lieb constrained - search functional given in eq . , the vignale rasolt
constrained - search functional in eq . is universal in the sense that it does not depend on the potential @xmath37 , only on the density @xmath38 .
another important characterization of the lieb functional is its convexity . to examine the convexity of @xmath98 ,
let @xmath99 and @xmath100 be arbitrary ( lebesque integrable ) functions and let @xmath101 , @xmath102 .
we then obtain @xmath103 \\ & \leq \inf_{\substack{\gamma_1\mapsto(\rho_1,\mathbf{j}_{\text{p}1})\\\gamma_2\mapsto(\rho_2,\mathbf{j}_{\text{p}2 } ) } } { \mathrm{tr}({(\lambda\gamma_1 + \mu\gamma_2 ) ( \tfrac{1}{2 } p^2 + w ) } ) } \\ & = \lambda f_{\text{vr}}[\rho_1 , \mathbf{j}_{\text{p}1 } ] + \mu f_{\text{vr}}[\rho_2,\mathbf{j}_{\text{p}2 } ] , \end{split}\ ] ] demonstrating that @xmath98 is convex in @xmath38 .
the key point in establishing the inequality above is to restrict the infimum over all density matrices @xmath104 to an infimum over all matrices of the form @xmath105 where @xmath106 and @xmath107 , thereby overestimating the infimum . given that the vignale rasolt functional is convex , it is uniquely represented by a convex conjugate functional . for the lieb functional @xmath26 $ ]
, the conjugate is the concave ground - state energy @xmath29 $ ] .
however , since @xmath76 $ ] is not concave , it can not be the conjugate to @xmath108 $ ] .
in the following , we identify the energy conjugate to @xmath109 $ ] , thereby arriving at a legendre fenchel formulation of cdft . inspection of the general expression in eq . and
the harmonic - oscillator example suggest the introduction of a new basic scalar potential , @xmath110 the choice of @xmath111 as basic potentials and @xmath38 as basic densities results in a theory where the hk variation principle takes the form of a legendre fenchel transformation with a linear pairing @xmath112 of the densities and potentials : @xmath113 & = \inf_{\gamma } { \mathrm{tr}({\gamma \bar{h}[u , \mathbf{a } ] } ) } \\ & = \inf_{\rho,\mathbf{j}_{\text{p } } } \left[f_{\text{vr}}[\rho,\mathbf{j}_{\text{p } } ] + ( \rho|u ) + ( \mathbf{j}_{\text{p}}|\mathbf{a } ) \right ] .
\end{split}\ ] ] here we have introduced the notation @xmath114 & = h[u - \tfrac{1}{2 } a^2 , \mathbf{a } ] , \\
\bar{e}[u,\mathbf{a } ] & = e[u - \tfrac{1}{2 } a^2 , \mathbf{a } ] .\end{aligned}\ ] ] the energy @xmath115 $ ] is now by construction concave , allowing it to be generated from the convex intrinsic energy @xmath116 $ ] by a reverse legendre
fenchel transformation : @xmath117 = \sup_{u,\mathbf{a } } \left [ \bar{e}[u,\mathbf{a } ] - ( \rho|u ) - ( \mathbf{j}_{\text{p}}|\mathbf{a } ) \right].\ ] ] thus , by a change of variables from @xmath15 to @xmath118 , we have restored the conjugate relation between the extrinsic and intrinsic energies characteristic of standard dft . strictly speaking , for @xmath116 $ ] and @xmath115 $ ] to form a conjugate pair
, we must specify their domains in the form of a banach space @xmath119 and its dual @xmath120 , respectively .
furthermore , we must demonstrate lower and upper semi - continuity of the intrinsic and extrinsic energies , respectively .
however , regarding the domains , we note here that , in lieb s formulation of standard dft , the vector space @xmath35 of potentials does not contain every potential with a square integrable ground state ( e.g. , it does not contain harmonic potentials ) , but @xmath121 does contain all coulomb potentials , thus covering most systems of interest .
a similar compromise is expected for cdft : we can not expect to identify a vector space @xmath119 of densities such that its dual @xmath120 includes all ground - state potentials @xmath16 . in particular ,
since gauge transformations may produce potentials @xmath122 that are arbitrarily ill - behaved at infinity , we expect some gauge restriction to be necessary .
in general , the functionals @xmath123 and @xmath98 are not differentiable .
consequently , we can not characterize the ground - state densities in the cdft hk variation principle of eq . in terms of functional derivatives .
on the other hand , in convex analysis , the proper tool for characterizing minimizers and maximizers are sub- and supergradients , respectively .
we here introduce sub- and supergradients in the context of cdft .
we begin by noting that an immediate consequence of the cdft variation principles in eqs . and
fenchel s inequality _ , @xmath124 \leq f_{\text{vr}}[\rho,\mathbf{j}_{\text{p } } ] + ( \rho|u ) + ( \mathbf{j}_{\text{p}}|\mathbf{a } ) , \label{eqfenchelineq}\ ] ] valid for any choice of potential @xmath125 and density @xmath126 .
moreover , equality holds if and only if @xmath127 is a ground - state density belonging to @xmath111 . to characterize ground - state densities and their potentials mathematically
, we use the concepts of subgradients and subdifferentials .
subgradient _ of @xmath98 at @xmath128 is an external potential @xmath129 for which the inequality @xmath96 \geq f_{\text{vr}}[\rho_0,\mathbf{j}_{\text{p}0 } ] - ( \rho-\rho_0|u_0 ) - ( \mathbf{j}_{\text{p}}-\mathbf{j}_{\text{p}0}|\mathbf{a}_0)\ ] ] holds for all @xmath126 ; see fig.[fig : subdiff ] .
clearly , all potentials @xmath130 for which the density @xmath131 is a minimizer in eq .
are subgradients of @xmath132 at @xmath131 .
the set of all subgradients at @xmath131 is known as the _ subdifferential _ of @xmath98 at @xmath133 and is denoted by @xmath134\subset x_{\text{p}}^*$ ] .
hence , to within a minus sign , the subdifferential at @xmath135 is the collection of all external potentials that have the same ground - state density @xmath135 .
$ ] : the set of all slopes ( potentials @xmath136 ) of just - touching tangent planes entirely below the graph at @xmath131 .
one particular tangent , with slope @xmath130 , is shown , while others are indicated by dashed lines .
a similar illustration can be made of the superdifferential @xmath137 $ ] .
[ fig : subdiff ] ] analogously , we consider the concave energy functional @xmath115 $ ] and its supergradients .
in general , @xmath138 is a _ supergradient _ of @xmath123 at @xmath139 if and only if @xmath140 is a subgradient of the convex functional @xmath141 at @xmath142 .
hence , @xmath131 is a supergradient of @xmath123 at @xmath142 if the inequality @xmath124 \leq \bar{e}[u_0,\mathbf{a}_0 ] + ( \rho_0|u - u_0 ) + ( \mathbf{j}_{\text{p}0}|\mathbf{a}-\mathbf{a}_0)\ ] ] holds for all @xmath143 .
this condition is satisfied precisely when @xmath131 is the density arising from a ( possibly degenerate ) ground state of @xmath144 $ ] .
the _ superdifferential _
@xmath145 $ ] is the collection of all supergradients of @xmath123 at @xmath142 . for all ground - state densities @xmath131 and associated potentials @xmath142
, we now have the following stationary conditions of the hk and lieb variation principles in eqs . and , respectively : @xmath146 , \quad & ( \rho_0,\mathbf{j}_{\text{p}0 } ) & \in \mathcal{b}_n^{\text{p } } \cap x,\\ ( \rho_0,\mathbf{j}_{\text{p}0 } ) & \in { \bar{\partial}}\bar{e}[u_0,\mathbf{a}_0 ] , \quad & ( u_0,\mathbf{a}_{0 } ) & \in \mathcal{u}_n \cap x^\ast.\end{aligned}\]]importantly , these conditions are equivalent : @xmath130 is a subgradient of @xmath98 at @xmath131 if and only if @xmath131 is a supergradient of @xmath123 at @xmath142 . hence , instead of a one - to - one mapping between individual potentials and individual ground - state densities , the convexity and concavity of the intrinsic and extrinsic energies , respectively , establish a mapping between the convex sets @xmath147 of degenerate ground - state densities and convex sets @xmath148 of potentials that give rise to identical ground - state densities ; see fig . [
fig : subdiff2 ] .
the sub- and superdifferentials are empty when no minimizer and maximizer exist in the corresponding optimization problems in eqs . and .
however , it is a general result of convex analysis that the subgradients ( supergradients ) of a convex ( concave ) function ( under certain semicontinuity conditions ) exist at a dense subset of the domain of the function . in cdft
, this result implies that the set of ground - state densities @xmath149 is dense in the set of all densities and that the set of potentials @xmath142 that support a ground state is dense in the set of potentials . . instead of a one - to - one mapping of individual potential pairs and corresponding ground - state density pairs ,
the sub- and superdifferentials of @xmath98 and @xmath123 , respectively , maps ( ground - state ) densities into convex sets of ( ground - state ) potentials and potentials into convex sets of densities.[fig : subdiff2 ] ] consider now the case of degenerate ground - state densities in eq . .
for a finite degeneracy @xmath150 , the superdifferential of the ground - state energy in the external potential @xmath142 is then a @xmath151-dimensional simplex with @xmath150 pure - state densities @xmath152 at the vertices : @xmath153 = \operatorname{co}\{(\rho_0^i , \mathbf{j}_{\text{p},0}^i ) , \ , | \
, i = 1 , g_\text d\ } , \label{corho}\ ] ] see fig .
[ fig : convex - hull ] for an illustration .
consequently , each ground - state density @xmath133 may be written as convex combination of the @xmath154 pure - state densities , @xmath155 as is well known , such degeneracies are either accidental or caused by symmetries of the hamiltonian .
$ ] of ground state densities for potentials @xmath111 with degeneracy @xmath156 .
this is a convex set , a three - dimensional simplex , with vertices at @xmath157 , i.e. , @xmath158 = \operatorname{co}\{(\rho_0^i,\mathbf{j}_{\text{p},0}^i)\}.$ ] in this example , we embed the simplex in @xmath159 , but in reality it is a three - dimensional subset of the density space @xmath119 .
the corresponding set of potentials is also a convex set , but usually with a more complicated structure .
see also fig .
[ fig : subdiff2 ] .
[ fig : convex - hull ] ] consider next a ground - state density @xmath133 with several maximizing potentials in the lieb variation principle in eq . .
like the superdifferential of the ground - state energy , the subdifferential of the vignale
rasolt density functional is a convex set .
let now @xmath142 be a potential with ground - state density @xmath160 and consider the family of external potentials @xmath161 for some potential @xmath162 and @xmath163 .
the associated hamiltonians are given by @xmath164 & = \bar{h}[u_0,\mathbf{a}_0 ] + \lambda k , \\ k & = \sum\nolimits_i [ \delta u_0(\mathbf{r}_i ) + \frac{1}{2 } \{\mathbf{p}_i , \delta \mathbf{a}_0(\mathbf{r}_i)\ } ] .\end{aligned}\ ] ] where @xmath165 denotes the anti - commutator .
clearly , if @xmath144 $ ] and @xmath166 commute , @xmath167 , k \right ] = 0 , \label{huk}\ ] ] then @xmath168 $ ] has the same eigenstates for all @xmath169 .
in particular , as @xmath169 changes continuously from @xmath170 , the external potentials in eq . will have the same ground - state density @xmath131 until a level crossing with the ground state occurs for @xmath168 $ ] , see capelle and vignale @xcite .
we note that the commutation condition in eq .
is sufficient but not necessary for the existence of degenerate maximizing potentials
. there may be several , independent perturbing potentials that commute with the reference hamiltonian @xmath144 $ ] . assuming a finite degeneracy @xmath171 of these potentials , we may write the subdifferential of the vignale
rasolt density functional as a @xmath172-dimensional simplex @xmath173 = -\operatorname{co}\{(u_0^i , \mathbf{a}_{0}^i ) , \,|\ , i = 1 , g_{\text{p}}\}.\ ] ] all potentials with the ground - state density @xmath160 may then be written as a convex combination , @xmath174 of the @xmath171 vertex potentials @xmath175 .
consider now the special case of dft , for which the hk and lieb variation principles are given by eqs . and , where @xmath29 $ ] and @xmath26 $ ] depend only on the scalar potential and the electron density , respectively .
let @xmath176 be a potential with ground - state density @xmath177 . clearly , the only scalar potentials @xmath4 that commute with the hamiltonian @xmath178 $ ] are the constant potentials @xmath49 .
it follows that the subdifferential of @xmath23 at @xmath177 is the convex set @xmath179 = - \ { v_0(\mathbf{r } ) + c \,|\ , c \in \mathbb r \},\ ] ] which may be regarded as a one - dimensional simplex with vertices @xmath180 . in dft
, therefore , the ground - state density @xmath177 determines the external potential @xmath181 uniquely up to an additive constant @xmath49 , in accordance with the hk theorem .
returning to cdft , consider next two potentials @xmath182 and @xmath183 with the same ground - state density @xmath131 . by the convexity of the subgradient of @xmath98 ,
all convex combinations @xmath184 then have the same ground - state density . recalling that @xmath185 , @xmath186 , and @xmath187 , the characterization of the non - uniqueness given in eq .
can be expressed in terms of the ordinary scalar potential @xmath15 .
if @xmath188 and @xmath189 give rise to the same density , then so do all potentials of the form @xmath190 with @xmath191 .
however , this set is not a convex set and not a subdifferential , due to the use of the @xmath37 rather than @xmath192 variables .
an advantage of the formulation of stationary conditions in cdft in terms of sub- and superdifferentials is that differentiability is not required . in general
, a sufficient condition for differentiability of a function at a point is that the function is continuous at this point and has a single sub- or supergradient there ; in the absence of continuity , differentiability is not guaranteed .
the ground - state energy is differentiable at all potentials @xmath111 that have a nondegenerate ground - state density , whereas the vignale rasolt functional is in principle nowhere differentiable since we may always add a constant term @xmath49 to the potential without affecting the ground - state densities .
however , assuming that this is the only cause of nondifferentiability of the potentials , we may in the absence of other degeneracies write @xmath193}{\delta \rho(\mathbf{r } ) } = -u(\mathbf{r } ) - c , \quad \frac{\delta f_{\text{vr}}[\rho,\mathbf{j}_{\text{p}}]}{\delta \mathbf{j}_{\text{p}}(\mathbf{r } ) } = -\mathbf{a}(\mathbf{r } ) , \label{statf}\ ] ] and @xmath194}{\delta u(\mathbf{r } ) } = \rho(\mathbf{r } ) , \quad \frac{\delta \bar{e}[u,\mathbf{a}]}{\delta \mathbf{a}(\mathbf{r } ) } = \mathbf{j}_{\text{p}}(\mathbf{r } ) , \label{state}\ ] ] where @xmath38 is the ( non - degenerate ) ground - state density of the potential @xmath111
. besides allowing formal application of theorems in convex analysis to cdft , the convex formulation given above has practical value .
after linear programming and optimization of quadratic functions , optimization of convex and concave functions is the mathematically most well - characterized type of optimization problem .
the fact that convex and concave optimization problems have a unique global optimum ( either in the form a single point or a convex set of optimal points ) and no additional local optima is of great value when devising practical optimization methods . in standard dft
, lieb s formulation @xcite of @xmath26 $ ] in terms of the legendre fenchel transform of the concave energy functional @xmath29
$ ] has proven useful in the study of functionals of interest in kohn sham theory @xcite .
in particular , the modulation of the two - electron interaction operator @xmath196 by a parameter @xmath169 such that , @xmath197 allows us to represent the ground - state energy @xmath198 $ ] at interaction strength @xmath169 in terms of its conjugate functional @xmath199 $ ] .
the standard choice is @xmath200 , but other choices are possible .
if the density supplied to @xmath199 $ ] is held fixed at a physical density generated by an appropriate ( high - level , systematically improvable ) _ ab - initio _ quantum - chemical methodology and if the value of the interaction - strength parameter is varied from @xmath170 to @xmath201 , then the adiabatic connection @xcite between the kohn
sham ( @xmath202 ) and physical ( @xmath203 ) systems can be studied numerically .
such studies for atomic and molecular species @xcite can provide useful insight into the failings of standard density - functional approximations and provide data for the construction and evaluation of new forms , based on the modeling of the adiabatic connection @xcite . having established a convex formulation of cdft in sec .
[ sec : restoreconc ] , it is possible to calculate the adiabatic connection in cdft in a manner similar to that of standard dft .
details of the adiabatic connection for cdft have been presented previously by liu @xcite .
the ground - state energy functionals at interaction strength @xmath169 are given by @xmath204 & = \inf_{\gamma } { \mathrm{tr}({\gamma h_{\lambda}[v,\mathbf{a } ] } ) } , \\
\label{eqelambdau } \bar{e}_{\lambda}[u,\mathbf{a } ] & = e_{\lambda}[u-\tfrac{1}{2 } a^2,\mathbf{a}],\end{aligned}\ ] ] where we have introduced the hamiltonian @xmath205 = \frac{1}{2 } \sum_k \pi_k^2 + \sum_k v(\mathbf{r}_k ) + w_\lambda . \label{hamlambda}\ ] ] from the exact functional @xmath206 $ ] or an accurate approximation to it , the adiabatic connection may be studied in terms the corresponding vignale
rasolt functional @xmath207 = \sup_{u,\mathbf{a } } \left [ \bar{e}_{\lambda}[u,\mathbf{a } ] - ( \rho | u ) - ( \mathbf{j}_{\text{p } } | \mathbf{a } ) \right ] , \label{acf}\ ] ] which needs to be evaluated for a fixed density @xmath38 and different values of @xmath169 in the interval @xmath208 .
typically , the density is the ground - state density for some external potential @xmath209 at @xmath203 .
the optimization in eq .
is trivial for @xmath210 since the optimal potential is then @xmath211 ; for @xmath212 , the optimization is non - trivial . in particular , for @xmath202 , the optimal potential is the kohn
sham potential @xmath213 given by @xmath214 where the classical coulomb or hartree potential @xmath215 and the exchange correlation potential @xmath216 are the functional derivatives of the corresponding energy components as in standard kohn
sham dft .
the exchange
correlation contribution to the vector potential is defined as @xmath217}{\delta \mathbf{j}_{\text{p}}},\ ] ] where differentiability in relevant directions is assumed .
these scalar and vector potentials then enter the cdft kohn
sham equations @xcite , which may be re - written in terms of @xmath213 as @xmath218 \varphi_p = \varepsilon_p \varphi_p.\ ] ] if the spin dependent @xmath219 term is included in the hamiltonian of eq . with the modified interactions of eq .
, then similar arguments apply .
two - component spinors rather than one - particle orbitals then occur in the kohn
sham equations , allowing for a treatment of non - collinear magnetism . even in the absence of external magnetic fields , violations of non - interacting @xmath15-representability
that is , the existence of a ground - state density of the fully interacting hamiltonian @xmath220 $ ] that can not be reproduced by slater - determinantal ground - states of the non - interacting hamiltonian @xmath221$]have been shown to be common in two - electron systems @xcite . in general , an extended kohn sham formalism , allowing for an ensemble description and fractional occupation numbers , is therefore required in cdft as well as in standard dft . to facilitate the optimization of @xmath222 $ ] at a general interaction strength @xmath169 , we restrict our attention to classes of potentials that can be parameterized in a simple way for example , as linear combinations of basis functions . the most direct way to benefit from the concavity of @xmath115 $ ] is to parameterize @xmath223 rather than @xmath15 in the affine form @xmath224 the use of @xmath223 rather than @xmath15 eliminates the @xmath225 term and the associated quadratic dependence on @xmath226 , thereby simplifying the equations obtained upon substitution in eq . .
importantly , it also ensures that all stationary points are true global maxima . to perform optimizations similar to those in refs .
@xcite , all that remains is to be able to calculate the ground - state energy @xmath227 $ ] with sufficient accuracy and to choose appropriate basis functions @xmath228 and @xmath229 .
derivatives with respect to the expansion coefficients @xmath230 and @xmath226 may then be used in a quasi - newton procedure analogous to that in ref .
the choice of basis functions and reference potentials raises the issue of a balanced descriptions of @xmath223 and @xmath44 , as well as the asymptotic limits and reduction of gauge freedom inherent in the finite basis set .
although important for practical implementation , these issues are beyond the scope of this article . the formulation of cdft with @xmath111 as the basic potential and with the intrinsic and extrinsic energies expressed as mutual legendre
fenchel transforms makes the introduction of spin straightforward .
the addition of the spin - zeeman operator to the hamiltonian introduces an energy term containing the spin density @xmath231 paired with the magnetic field , @xmath232 . a partial integration transfers the curl operator to the spin density and gives a surface term if the integration domain is finite .
neglect of the surface term leads to the energy term @xmath233 and a theory with @xmath234 as the basic density . here , @xmath235 is the sum of the paramagnetic current and the _ spin current _ @xmath236 .
all results above remain valid with @xmath237 replaced by @xmath238 and with suitable modifications of the hamiltonian and definitions of @xmath16 and @xmath239 .
the use of @xmath240 as a basic density has been discussed by capelle and gross @xcite as a way to translate functionals between spin - density functional theory ( sdft ) and cdft . the alternative formulation of current - spin dft ( csdft ) in terms of the charge density @xmath17 , the spin density @xmath241 , and the paramagnetic current density @xmath237 as separate basic variables is a less attractive formal theory since two independent potentials such as @xmath111 can not be conjugate to three independent densities @xmath242 . on the other hand , construction of practical approximate exchange
correlation functionals may be substantially more difficult with the basic variables @xmath17 and @xmath238 .
mathematically , it is not surprising that a theory formulated in terms of magnetic vector potentials and wave functions both gauge - dependent objects makes use of gauge - dependent basic variables . indeed , with gauge - dependent notions being so deeply entrenched in the theory , it is not trivial to construct a useful reformulation that features only gauge - invariant basic variables . on the other hand , from a physical point of view ,
it is somewhat unappealing that the paramagnetic current , rather than the physical current , arises as a basic density . therefore , some authors have attempted the construction of an alternative cdft , with the physical current as a basic variable . in particular ,
pan and sahni have gone far in arguing that the paramagnetic current density , in some sense , can not correctly be regarded as a basic cdft variable @xcite .
an important question that arises in cdft is to what extent an hk theorem is possible when the physical densities @xmath243 are chosen as the basic densities that is , whether @xmath244 always implies that @xmath245 .
two simple observations lend plausibility to this claim .
first , for one - electron systems , @xmath37 can be found explicitly given @xmath243 .
writing the one - electron wave function as @xmath246 where @xmath247 and @xmath248 are real valued , we obtain the density @xmath249 and physical current density @xmath250 .
because of the identity @xmath251 , the physical densities @xmath17 and @xmath252 determine the external magnetic field , @xmath253 the scalar potential @xmath15 may be determined up to a constant from the eigenvalue equation @xmath254 and the observation that @xmath255 , yielding @xmath256 \rho^{1/2}(\mathbf{r } ) = 0,\ ] ] from which @xmath257 is uniquely determined .
hence , for a one - electron system , @xmath80 and @xmath15 are both determined by @xmath17 and @xmath252 . for an @xmath0-electron system ,
we obtain more generally @xmath258 where @xmath259 is the vorticity introduced by vignale and rasolt @xcite .
vorticity is a gauge - invariant quantity , but it is not clear whether it can be uniquely reconstructed from @xmath243 , without a priori knowledge of the external magnetic field @xmath80 .
following pan and sahni @xcite , the second observation is that , while there is no hk theorem for cdft in terms of _ paramagnetic _ densities , the counterexamples to the existence of an hk theorem for paramagnetic densities ( such as the harmonic - oscillator system in sec .
[ sec : electronic ] ) do not preclude an hk theorem for physical densities . to see this , note that these counterexamples all exhibit different potentials @xmath260 with the same ground state @xmath261 and therefore the same paramagnetic density @xmath262 .
however , according to the original hk theorem , this situation is impossible when @xmath263 since this would imply @xmath264 and therefore @xmath265 .
we must therefore assume that @xmath266 in the counterexamples .
however , from the assumption @xmath267 , it then follows that @xmath268 . in short ,
if @xmath260 share the same ground - state wave function , then the _ physical densities _ are not the same .
it remains to explore whether it is possible to have different ground states @xmath269 but the same physical densities @xmath270 ; if possible , then no hk theorem exists for the physical current densities .
general arguments for an hk theorem for physical current densities have been put forth by pan and sahni @xcite and by diener @xcite . however , as discussed below , neither of these arguments amounts to a rigorous proof . to our knowledge ,
the existence of an hk theorem for physical current densities therefore remains open .
the standard hk theorem of dft states that @xmath271 implies @xmath272 .
the proof has two parts : first , it is shown that scalar potentials that differ by more than a constant must have different wave functions @xmath273 and @xmath274 ; second , this result is combined with the rayleigh ritz variation principle to show that the assumptions @xmath275 and @xmath276 lead to a contradiction .
it follows that @xmath277 implies @xmath278 . for cdft with physical densities ,
a proof along the same lines has been attempted by pan and sahni @xcite .
the first part of their argument establishes that different potentials @xmath279 can not yield the same physical density @xmath280 if the ground - state wave functions are the same .
that is , without loss of generality , it may be assumed in an hk - type argument that the wave functions are different .
the second part of their argument seeks to establish , by the rayleigh
ritz variation principle , that gauge - inequivalent potentials can not have both different ground - state wave functions and the same physical ground - state density . to this end , two potentials @xmath281 , @xmath282 , are considered and it is argued that a contradiction arises from the following assumptions : * the potentials have the same physical density @xmath283 . *
the potentials differ by more than a gauge transformation , @xmath284 .
no contradiction can result from ( a ) alone , so the assumption ( b ) is crucial for any reductio ad absurdum argument to succeed .
for example , because of gauge freedom , ( a ) and the negation of ( b ) are certainly not contradictory since this corresponds to the perfectly consistent situation where two potentials that differ by a gauge transformation give rise to the same physical density .
a correct proof must therefore contain at least one step that makes use of ( b ) .
however , while stated as an assumption , ( b ) is in fact never used in pan and sahni s argument .
the argument must consequently be invalid . in more detail , we here identify an erroneous step in the reasoning in ref .
@xcite , in particular eqs .
( 35)(40 ) . with @xmath273 and @xmath274
denoting the ground states of the two potentials , we obtain eq .
( 35 ) from ref .
@xcite : @xmath285 { | { \psi } \rangle } < { \langle { \psi ' } | } h[v,\mathbf{a } ] { | { \psi ' } \rangle}.\ ] ] the identity @xmath286 = h[v',\mathbf{a } ' ] + ( h[v,\mathbf{a } ] - h[v',\mathbf{a}'])$ ] yields for the expectation value on the right - hand side
@xmath287 { | { \psi ' } \rangle } & = e ' + { \langle { \psi ' } | } \tfrac{1}{2 } \pi^2 + v - \tfrac{1}{2 } \pi'^2 - v ' { | { \psi ' } \rangle } \\ & = e ' + { \langle { \psi ' } | } \tfrac{1}{2 } \{\boldsymbol{\pi},\mathbf{a}\ } + v - \tfrac{1}{2 } a^2 - \tfrac{1}{2 } \{\boldsymbol{\pi}',\mathbf{a}'\ } - v ' + \tfrac{1}{2 } a'^2 { | { \psi ' } \rangle}. \label{hexp } \end{split}\ ] ] it is here important to distinguish between @xmath288 and @xmath289 since the representation of the mechanical momentum operator is gauge and vector - potential dependent . to proceed ,
we explicitly write out the corresponding physical current density operators , which for the two potentials are given by @xmath290 from these , we may calculate the physical current ( assumed to be the same in the two cases ) and its interaction with some vector potential @xmath291 as @xmath292 where it is important to use primed or unprimed quantities consistently . to ensure that we are using the correct current - density operator for @xmath274 in eq .
, we insert the identity @xmath293 yielding , @xmath294 where we have introduced @xmath295 and @xmath296 and also used the inequality in eq . .
carrying out the above argument with primed and unprimed variables interchanged , we obtain the strict inequality @xmath297 where the last term does not vanish since @xmath298 is symmetric in the primed and unprimed variables . in agreement with the hk theorem
, a contradiction arises if ( and only if ) @xmath299 .
when @xmath41 and @xmath300 differ where @xmath301 ( as a result of different physical fields or by a gauge transformation ) , no contradiction arises . in the argument given in ref .
@xcite , the authors incorrectly identify @xmath302 and @xmath303 , leading to their eq .
( 38 ) , which differs from eq . above by the replacement of @xmath304 with @xmath305 .
since their term is antisymmetric rather than symmetric in the primed and unprimed variables , the authors obtain a contradiction @xmath306 irrespective of @xmath44 and @xmath307 , leading to the unjustified conclusion that ( @xmath308 determines @xmath37
. an earlier attempt to prove an hk - type theorem for physical currents by diener @xcite invokes an intriguing strategy for eliminating the term @xmath309 from eq . .
the key idea is to replace the external vector potential @xmath44 by an effective vector potential @xmath310 $ ] defined to reproduce a given physical current density @xmath252 from a given wave function @xmath273 so that @xmath311.\ ] ] consider now the following rearrangement of the kinetic energy term @xmath312 by definition , @xmath313,\mathbf{a}\ } { | { \psi } \rangle } = ( \mathbf{j}|\mathbf{a})$ ] . for a prescribed physical current density @xmath314 , the expectation value of an arbitrary wave function is then @xmath315 { | { \psi } \rangle } & = { \langle { \psi } | } \tfrac{1}{2 } ( p^2 - a_{\text{eff}}^2 ) + w { | { \psi } \rangle } \nonumber \\ & + ( \mathbf{j}_0|\mathbf{a } ) + ( \rho|v ) + \tfrac{1}{2}(\rho | ( \mathbf{a}_{\text{eff}}-\mathbf{a})^2 ) .
\label{eqhwitheffa}\end{aligned}\ ] ] the total current evaluated using the effective momentum operator @xmath316 instead of the true mechanical momentum operator @xmath317 always evaluates to the prescribed current @xmath314 .
diener then defines a universal density functional of the form @xmath318 & = \inf_{\psi\mapsto\rho } { \langle { \psi } | } h_{\text{eff}}[\mathbf{j},\psi ] { | { \psi } \rangle } , \\
h_{\text{eff}}[\mathbf{j},\psi ] & = \tfrac{1}{2 } p^2 - \tfrac{1}{2 } a_{\text{eff}}[\mathbf{j},\psi]^2 + w. \label{dienerh}\end{aligned}\ ] ] by exploiting this functional , diener derives the inequality @xmath319 { | { \psi_0 } \rangle } \leq { \langle { \psi'_0 } | } h[v,\mathbf{a } ] { | { \psi'_0 } \rangle } - \frac{1}{2 } ( \rho_0 | \delta a^2),\ ] ] in lieu of the usual strict rayleigh ritz inequality underlying the hk proof as given in eq . .
when @xmath320 is the ground state of potentials that differ by more than a gauge , @xmath321 , the above non - strict inequality is ( without further ado ) taken to be strict in diener s presentation .
if a strict inequality is accepted , a standard reductio ad absurdum proof is possible because the last term of eq
. cancels to yield the contradiction @xmath322 .
we now consider two technical problems not addressed by diener .
the first is that the expectation value being minimized in @xmath323 is not bounded from below for densities that vanish at some point .
although unusual , such ground - state densities can arise for small molecules in strong magnetic fields .
consider a wave function @xmath324 giving rise to a density @xmath177 that vanishes at some point in space @xmath325 .
let us also introduce spherical coordinates @xmath326 about @xmath325 .
then , for the wave functions @xmath327 with integer @xmath328 , we see that @xmath329 and @xmath324 give rise to the same density @xmath177 but to different paramagnetic current densities related by @xmath330 where @xmath331 is the unit vector in the direction specified by @xmath332 . as a result , @xmath333 = \mathbf{a}_{\text{eff}}[\mathbf{j } , \phi_0 ] - \frac{m \hat{\boldsymbol{\phi}}}{r \sin(\theta ) } = \mathbf{a}_0 - \frac{m \hat{\boldsymbol{\phi}}}{r \sin(\theta)}.\ ] ] calculating the expectation value of the effective hamiltonian in eq . , some terms arising from @xmath334 cancel terms arising from @xmath335 , leaving @xmath336 for physical currents @xmath252 with a nonzero last integral , for example a circular current , the expectation value can be decreased without bound , demonstrating that @xmath337 $ ] is not well defined on the full domain @xmath19 .
the second technical problem in the derivation of the inequality in eq .
concerns the strategy of choosing a prescribed @xmath314 such that the effective vector potential equals the external vector potential .
this amounts to a self - consistency condition [ eq . ( 11 ) of ref .
] , @xmath338 = \mathbf{j}_{\text{p};\psi[v,\mathbf{a } ] } + \rho_0 \mathbf{a},\ ] ] where @xmath339 minimizes @xmath340 { | { \phi_{\text{d } } } \rangle}$ ] and @xmath341 $ ] is the ground state of the external potentials giving rise to @xmath283 .
diener s derivation of the inequality in eq .
hinges on the ability to reproduce the paramagnetic current density from a ground state @xmath341 $ ] of @xmath286 $ ] by a minimizing wave function in @xmath342 $ ] . however , it is unclear whether this is always possible . in lieu of a rigorous proof for an hk - type theorem for physical currents ,
it is possible to proceed by conjecturing such a result and exploring its consequences . additionally , an important point is that a mapping from ground state densities to potentials may not be required for a formulation of cdft , provided that the theory can be constructed by some other means such as a constrained search or legendre fenchel transformation formalism .
we shall here explore such issues in the framework introduced by pan and sahni @xcite in more detail . a complication due to the choice of the physical current as
a basic variable is that constraints of the type `` @xmath343 '' require explicit reference to a vector potential , because the physical current is not determined by the wave function alone .
care must therefore be exercised when developing a constrained - search formalism for physical currents . for each magnetic field under consideration
, we fix a gauge .
hence , we choose a mapping @xmath344(\mathbf{r})\ ] ] from magnetic fields to magnetic vector potentials , and also fix the constant shift of scalar potentials in some way . from the conjecture
that a ground state density @xmath243 uniquely determines a gauge class of potentials @xmath345 , we may now write mappings @xmath346)\ ] ] hence , we may write the scalar potential , the external magnetic field and its vector potential as functionals @xmath347 $ ] , @xmath348 $ ] , and @xmath349 $ ] of the physical densities ; these are representatives of the equivalence class @xmath345
. given potentials @xmath37 , we may determine the corresponding ground state @xmath350 and physical ground - state densities @xmath351 , and express the energy as @xmath352 = { \langle { \psi_0 } | } \tfrac{1}{2 } p^2 + w { | { \psi_0 } \rangle } + ( \rho|v-\tfrac{1}{2 } a^2 ) + ( \mathbf{j}|\mathbf{a } ) .
\end{split}\ ] ] using the mapping from densities to potentials , a universal density functional could in principle internally reconstruct the ground state potentials , and obtain the intrinsic energy via @xmath353 & = e\left[w[\rho,\mathbf{j}],\mathbf{a}[\mathbf{b}[\rho,\mathbf{j}]]\right ] - ( \rho\,|\,w[\rho,\mathbf{j}]-\tfrac{1}{2 } a[\mathbf{b}[\rho,\mathbf{j}]]^2 ) \\ & \ \ - ( \mathbf{j}\,|\,\mathbf{a}[\mathbf{b}[\rho,\mathbf{j } ] ] ) .
\end{split}\ ] ] furthermore , we may define a non - universal constrained - search functional that depends explicitly on the external vector potential , @xmath354 & = \inf_{\gamma\mapsto(\rho,\mathbf{j}-\rho\mathbf{a } ) } { \mathrm{tr}({\gamma ( \tfrac{1}{2 } p^2 + w ) } ) } \\ & = f_{\text{vr}}[\rho,\mathbf{j}-\rho\mathbf{a } ] , \end{split}\ ] ] where we note that @xmath355 . in principle , given the conjectured one - to - one mapping between potentials and physical densities , a functional of physical densities may internally reconstruct the corresponding potentials .
this observation is relied on in the definition of @xmath356 and it is tempting to consider also a universal constrained - search functional that exploits this idea , @xmath357 & = \inf_{\gamma\mapsto(\rho,\mathbf{j}-\rho\mathbf{a}[\mathbf{b}[\rho,\mathbf{j } ] ] ) } { \mathrm{tr}({\gamma ( \tfrac{1}{2 } p^2 + w ) } ) } \\ & = f_{\text{vr}}[\rho,\mathbf{j}-\rho\mathbf{a}[\mathbf{b}[\rho,\mathbf{j } ] ] ] .
\end{split}\ ] ] consider now the reformulation of the minimization in eq . as a nested minimization over physical densities and wave functions .
@xmath358 & = \inf_{\rho',\mathbf{j } ' } \big [ \overbrace{\inf_{\gamma\mapsto\rho',\mathbf{j}'-\rho'\mathbf{a } } { \mathrm{tr}({\gamma ( \tfrac{1}{2 } p^2 + w)})}}^{f^{\text{cs1}}_{\text{ps}}[\rho',\mathbf{j}',\mathbf{a } ] } \\ & \ \ + ( \rho'|v-\tfrac{1}{2 } a^2 ) + ( \mathbf{j}'|\mathbf{a } ) \big ] \end{split}\ ] ] a constrained - search - like expression is thus possible , but the functional that arises from the nested minimization is the non - universal functional @xmath359 .
a minimization over the universal functional @xmath360 is here not equivalent , @xmath361 & \neq \inf_{\rho',\mathbf{j } ' } \big [ \overbrace { \inf_{\gamma\mapsto\rho',\mathbf{j}'-\rho'\mathbf{a}[\mathbf{b}[\rho',\mathbf{j } ' ] ] } { \mathrm{tr}({\gamma ( \tfrac{1}{2 } p^2 + w)})}}^{f^{\text{cs2}}_{\text{ps}}[\rho',\mathbf{j } ' ] } \\ & \ \ + ( \rho'|v-\tfrac{1}{2 } a^2 ) + ( \mathbf{j}'|\mathbf{a } ) \big ] .
\end{split}\ ] ] to establish the non - equality , consider the introduction of a new scalar potential @xmath362 . written in terms of @xmath363 , the right - hand side of eq . has the form of a legendre fenchel transformation and consequently a functional of @xmath364 that is concave by construction .
however , writing the harmonic oscillator energy of sec .
[ sec : restoreconc ] in terms of this scalar potential , we see that the resulting energy functional is not concave in @xmath364 ( it is in fact only concave in scalar potentials @xmath365 for @xmath366 ) .
hence this non - concave energy functional can not equal the concave functional defined in terms of @xmath360 . in previous work
( see , e.g. , eqs .
( 24)(28 ) of ref .
@xcite ) , these two non - equivalent functionals were conflated .
the constrained - search approach to cdft with the physical current as a basic variable is substantially complicated by the fact that a wave function does not determine the physical current .
it does not appear possible to construct a functional that is both universal and admits a straightforward variation principle .
moreover , unlike the situation when @xmath237 is a basic variable , a simple redefinition of the scalar potential does not simultaneously yield a concave energy functional and linear pairing between potentials and physical densities .
an avenue for further study could be to consider the properties of a universal intrinsic energy functional of the physical densities .
such a functional can not be convex in the pair @xmath243 as a whole , but could possibly be convex in @xmath17 ( for a fixed @xmath252 ) and concave in @xmath252 ( for a fixed @xmath17 ) , which would enable at least partial legendre fenchel transformations .
in this work we have compared formulations of cdft based on different choices of basic variables .
while the usual formulation in terms of @xmath367 and @xmath37 does not lead to a one - to - one mapping between potentials and densities , it allows for the construction of a useful density functional theory via a constrained search approach .
a drawback with this choice of basic variables is that @xmath76 $ ] is not concave , making a full legendre fenchel transform treatment infeasible .
however , as shown here , concavity can be restored by considering the conjugate variables @xmath368 and @xmath111 , where @xmath369 .
this allows the application of convex analysis in analogy with lieb s formulation of standard dft @xcite .
such a formulation is particularly natural in the context of the study of adiabatic connections for cdft functional construction , which can be done in a manner similar to that undertaken for standard dft @xcite .
the information garnered from such analysis would be suitable for comparison with kohn
sham implementations of cdft based on functionals of @xmath38 .
alternative cdft formulations based on the physical densities @xmath243 have also been critically examined .
pan and sahni s @xcite recent attempt to formulate such a theory is found to be unsatisfactory .
both their attempt to prove the existence of a one - to - one mapping between potentials and densities and their constrained - search formulation were found to be flawed .
an earlier attempt to formulate a cdft in terms of the physical current by diener @xcite has also been examined , and technical difficulties with this approach have been highlighted . despite the appealing physical motivation behind this choice of basic variables , we thus find that a formal justification for such a framework is currently lacking .
furthermore , while it remains open whether or not an analogue of the hohenberg
kohn theorem holds for the physical current , other aspects of standard dft such as the variation principle , the constrained - search formalism , and formulations in terms of legendre fenchel transformations do not straightforwardly carry over to this type of cdft .
we conclude that the most common formulation in terms of @xmath367 is presently the most convenient and viable formulation of cdft .
let @xmath31 be a normed vector space and @xmath121 its dual that is , the the linear space of all continuous linear functionals on @xmath31 .
a function @xmath370 is said to be convex if it satisfies the relation @xmath371 for all @xmath372 .
the effective domain , @xmath373 , is the set of @xmath374 for which @xmath375 .
a function @xmath376 is lower semi - continuous at @xmath377 if , for any @xmath378 , there exists @xmath379 such that @xmath380 whenever @xmath381 ; @xmath376 is lower semi - continuous if it is lower semi - continuous at all @xmath374 .
a lower semi - continuous convex function @xmath370 may be represented by its conjugate @xmath382 in the manner @xmath383 , \\
f(x ) & = \sup_{y \in x^\ast } [ ( x | y ) - f^\ast(y)],\end{aligned}\ ] ] where @xmath384 is also lower semi - continuous and convex .
the dual function @xmath385 is said to be a subgradient of @xmath376 at @xmath386 if it satisfies the inequality @xmath387 the set of all subgradients of @xmath376 at @xmath377 is called the subdifferential @xmath388 and is a ( possibly empty ) convex subset of @xmath121 .
subgradients and subdifferentials of @xmath384 are defined in an analogous manner .
the function @xmath376 and its conjugate satisfy fenchel s inequality @xmath389 , which is sharpened into the equality @xmath390 whenever the equivalent reciprocal relations @xmath391 are satisfied .
a function @xmath392 is said to be concave if @xmath393 is convex .
its conjugate is defined in the same manner as for convex functions but with @xmath394 replaced by @xmath395 ; likewise , subgradients and subdifferentials are defined as for a convex function but with the inequality sign reversed in eq . .
this work was supported by the norwegian research council through the coe centre for theoretical and computational chemistry ( ctcc ) grant no .
179568/v30 and the grant no .
171185/v30 and through the european research council under the european union seventh framework program through the advanced grant abacus , erc grant agreement no . 267683 .
a. m. t. is also grateful for support from the royal society university research fellowship scheme .
52ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrev.136.b864 [ * * , ( ) ] link:\doibase 10.1073/pnas.76.12.6062 [ * * , ( ) ] link:\doibase 10.1002/qua.560240302 [ * * , ( ) ] @noop _
_ , ed .
( , ) link:\doibase 10.1016/j.theochem.2006.05.012 [ * * , ( ) ] link:\doibase 10.1103/physreva.73.012513 [ * * , ( ) ] link:\doibase 10.1088/0305 - 4608/4/8/013 [ * * , ( ) ] link:\doibase 10.1016/0038 - 1098(75)90618 - 3 [ * * , ( ) ] link:\doibase 10.1103/physrevb.13.4274 [ * * , ( ) ] link:\doibase 10.1103/physrevb.15.6006.3 [ * * , ( ) ] link:\doibase 10.1103/physrevb.15.2884 [ * * , ( ) ] link:\doibase 10.1063/1.478234 [ * * , ( ) ] link:\doibase 10.1063/1.1535422 [ * * , ( ) ] link:\doibase 10.1063/1.3082285 [ * * , ( ) ] link:\doibase 10.1063/1.3380834 [ * * , ( ) ] link:\doibase 10.1063/1.3488100 [ * * , ( ) ] link:\doibase 10.1063/1.3660357 [ * * , ( ) ] link:\doibase 10.1103/physreva.50.3089 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.69.035113 [ * * , ( ) ] link:\doibase 10.1103/physreva.80.032510 [ * * , ( ) ] link:\doibase 10.1002/qua.22862 [ * * , ( ) ] link:\doibase 10.1016/j.jpcs.2011.12.023 [ * * , ( ) ] link:\doibase 10.1103/physreva.85.052502 [ * * , ( ) ] link:\doibase 10.1103/physreva.86.042502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.59.2360 [ * * , ( ) ] link:\doibase 10.1103/physrevb.37.10685 [ * * , ( ) ] link:\doibase 10.1103/physreva.54.1328 [ * * , ( ) ] link:\doibase 10.1103/physreva.53.r5 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreva.59.209 [ * * , ( ) ] link:\doibase 10.1103/physreva.74.062511 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.245106 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.036403 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreva.38.1149 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.65.113106 [ * * , ( ) ] link:\doibase
10.1088/0953 - 8984/3/47/014 [ * * , ( ) ] @noop _ _
( , ) link:\doibase 10.1063/1.464304 [ * * , ( ) ] link:\doibase 10.1103/physreva.59.51 [ * * , ( ) ] link:\doibase 10.1103/physreva.62.012502 [ * * , ( ) ] link:\doibase 10.1021/ct8005248 [ * * , ( ) ] link:\doibase 10.1063/1.2749510 [ * * , ( ) ] link:\doibase 10.1103/physreva.80.022517 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.78.1872 [ * * , ( ) ] | the selection of basic variables in current - density functional theory and formal properties of the resulting formulations are critically examined .
focus is placed on the extent to which the hohenberg
kohn theorem , constrained - search approach and lieb s formulation ( in terms of convex and concave conjugation ) of standard density - functional theory can be generalized to provide foundations for current - density functional theory .
for the well - known case with the gauge - dependent paramagnetic current density as a basic variable , we find that the resulting total energy functional is not concave .
it is shown that a simple redefinition of the scalar potential restores concavity and enables the application of convex analysis and convex / concave conjugation . as a result ,
the solution sets arising in potential - optimization problems can be given a simple characterization .
we also review attempts to establish theories with the physical current density as a basic variable . despite the appealing physical motivation behind this choice of basic variables , we find that the mathematical foundations of the theories proposed to date are unsatisfactory .
moreover , the analogy to standard density - functional theory is substantially weaker as neither the constrained - search approach nor the convex analysis framework carry over to a theory making use of the physical current density . |
my interest in the history of the aharonov - bohm effect [ ab ] started when i joined birkbeck college in 1961 where two of key players in the discovery held professorships .
werner ehrenberg held the chair of experimental physics and was head of the department , while bohm had just been appointed to a chair in theoretical physics .
my first venture into this effect resulted in some personal embarrassment .
i had decided to write a paper on the effect mainly to clarify my own understanding of the phenomenon .
when ehrenberg saw a copy of the paper he confronted me with a comment spoken with a strong german accent , ach hiley , zis ab effect that you are discussing , is it the one that siday and i discovered ?
" poor dear , gentle werner , sidelined in history by a young new member of his own staff !
it was my original intention to describe the way in which ehrenberg and siday @xcite discovered the effect in the 1940s .
i had at hand in the department a number of people who were working with ehrenberg and siday when they made their discovery , so i am able to obtain first hand accounts of how the story unfolded .
my only regret was that ray siday had passed on , but the first hand recollections of the man were still very much alive in the department .
the events leading up to the discovery took place in the late 1940s and it turns out to be a fascinating story .
i will only briefly comment on the re - discovery of the effect by yakir aharonov and david bohm @xcite in the late 1950s .
the brevity of my comments is no reflection on their work , which was very significant since it marked the beginning of gauge field theories @xcite . although i had many hours of discussion with david bohm , the @xmath0 effect was not high on the agenda , as we were interested in a wider range of physical and philosophical questions .
i will leave a discussion of the ab s papers and the subsequent to the development of general gauge theory to others .
as i was beginning to collect the background material , my attention was drawn to the abstract of a talk given by werner franz to a physical society meeting in danzig in 1939 @xcite .
it left me with the impression he might have been aware of the effect even at this early date .
could the talk he presented have contained a first mention of this effect and _ ipso facto _
could he have been the first to discover this effect ?
how does one investigate something that took place in the then free city of danzig in may 1939 , when it was just about to be involved in violent political events ? . by august
, the city had undergone a _ coup detat _ and later in september a german battleship used the harbour to open fire on the polish city of westerplatte .
not the atmosphere to think deeply about physics
! the starting point for this investigation , then , can only be the franz abstract .
here we reproduce it in its original form . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ item 5 .
w. franz ( knigsberg , pr ) : _ elektroneninterferenzen i m magnetfeld .
_ nach de broglie ist einem elektron vom impuls @xmath1 eine wellenlnge @xmath2 zugeordnet , worin @xmath3 die plancksche konstante .
i m magnetfeld tritt zum korpuskularen impulse @xmath4 der zusatz @xmath5 , wo @xmath6 das vektor potential des magnetfeldes .
da @xmath6 nur bis auf einen gradienten bestimmt ist , hat i m magnetfeld auch die wellenlnge keinen eindeutigen physikalischen sinn .
doch fllt diese vieldeutigkeit bei der physikalischen anwendung , der interferenz .
- der wellenzahlvektor eines eleklronenfeldes @xmath7 . , nach de broglie also @xmath8 ist der gradient der phase , muss also rot - frei sein .
i m magnetfeld ist nun rot @xmath9 ; der zusatz @xmath5 stellt die bedingung rot @xmath10 wieder her .
- bei der beugung am doppelspalt ergibt sich als bedingung fr ein interferenzmaximum @xmath11 worin @xmath12 die geometrische gangdifferenz der beiden mglichen korpuskularen bahnen und @xmath13 der magnetische fluss durch die von ihnen eingeschlossene flche ist @xmath14 eine ganze zahl .
hiernach ergibt sich in bereinstimmung mit der erfahrung , dass die interferenz nur durch die richtung bestimmt wird , in welcher die elektronen die spalte erreichen und verlassen .
die durch die bahnkrmmung i m magnetfeld hervorgerufenen gangdifferenzen werden durch den zusatz @xmath15 aufgehoben[multiblock footnote omitted ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ although franz clearly states that the interference depends on the magnetic flux enclosed by the electron paths , he adds that the path difference is caused by the curvature of the paths in the magnetic field " .
this last statement clouds the issue because it is unclear whether he realised that the electron paths need not experience any magnetic field at all and still produce the same interference shift provided the path encloses the field .
this is the key to the whole effect .
i originally thought that this was an abstract of a paper , but i failed to find any such paper .
it then transpired that it was an abstract of a seminar that franz gave to the physics meeting
gauvereins ostland in danzig " held from 18 to 29 may 1939 .
soon after this abstract came to my attention @xcite , i was approached by gottfried mllenstedt , an experimentalist who had pioneered some brilliant electron interference experiments @xcite , and asked to supply some biographical details of both ehrenberg and sidday .
he was preparing a paper on the history of the electron biprism @xcite , an instrument that he had perfected and had used in his electron of interference experiments . in this article ,
mllenstedt @xcite informs us that he attended the danzig meeting as a young physics student and it was at that meeting that he heard the term electron interferometry " for the first time . to recall the meeting he used a paper by franz @xcite , which was written in 1965 , and translates a key part of this paper which is reproduced here .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in presence of an electromagnetic field , the momentum of a particle with charge @xmath16 is known to be @xmath17 where @xmath18 is the vector potential from which the magnetic field strength is determined as @xmath19 .
the phase difference @xmath20 between two rays ( a ) and ( b ) connecting two points 1 and 2 is determined by @xmath21 introducing the expression for @xmath22 is from above , the term @xmath23 yields the same path difference as in absence of a magnetic field whereas , according to stokes theorem , the loop integral over @xmath18 may be transformed to a surface integral over @xmath24 , i.e. , the magnetic flux @xmath25 , yielding @xmath26 this simple relation which should be the first thing taught in a lecture on wave mechanics for beginners after introducing the magnetic field ( strangely enough i could not find it in any lecture notes except my own ) shows that the phase difference between electron rays depends on the magnetic flux included between the rays , even if the rays do not run in a magnetic field .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in 1939 , after the lecture by walter franz on , walther kossel discussed the possibility of an experimental proof , but he came to the conclusion that , at that time , an experimental proof was not feasible .
mllenstedt writes , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
i remember w. kossel saying i hear the message but i lack an electron interferometer . " _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ my reading of these last two sentences strongly suggests that there was a discussion about the ab effect at that danzig meeting , but it was brought to a end simply because no one in the group had the means to explore the effect experimentally . but did that mean that franz had really discovered the effect ?
in contrast to this , i have a letter from mllenstedt dated feb .
12 1993 , which is six years before mllenstedt s paper appeared @xcite . in it
mllenstedt writes _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ i am enclosing a copy of a recent paper on the amerigo effect " printed in
physikalische bltter " .
i think that possibly the ab effect is also some kind of an amerigo effect for which the credit should rather have been given to ehrenberg and siday .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the amerigo effect " was a phrase coined by mllenstedt and walther kossel @xcite to refer to the mis - attribution of the discoverer of some physical effect .
the word takes its meaning from amerigo vespucci who sailed west after columbus made his discovery of america .
the medici bank of florence , for whom amerigo worked as a local branch manager in seville , spotting an advertising opportunity , decided to give circulation in europe to an account of amerigo s journey .
a geographer waldseemller subsequently attributed the discovery of the new continent to amerigo !
unfortunately mllenstedt died in 1997 and by the time the history of the electron biprism " article reached me , it was too late to obtain a clarification of this point , so it is still an open question as to whether franz had spotted the effect .
werner ehrenberg was born in berlin and studied philosophy , physics and mathematics at the university of berlin and took his phd at heidelberg .
he was an assistant in the physical institute of the technische hochschule in stuttgart but was dismissed for being jewish in 1933 and sort refuge in the uk .
he worked at birkbeck under blackett on a grant from the academic assistance council , before working in industry during ww ii . at the end of the war he returned to birkbeck where he worked under j. d. bernal before becoming established in his own right , finally taking the chair of experimental physics at birkbeck .
his interests in physics were wide and varied .
his early interests were in x - rays and electron optics .
indeed this early work centred on the task of developing a method to produce soft focus x - rays that could be used on biological molecules @xcite , @xcite . with walter spear he developed and built the fine focus x - ray generator that maurice wilkins , rosalind franklin and ray gosling used to study the structure of dna .
it was this experimental work that enabled crick and watson to propose the double helix structure that won them , together with wilkins , the nobel prize .
ehrenberg s later interests were in electrical conduction in semiconductors and metals @xcite .
ray siday was a totally different character .
he took a first in the b.sc .
special physics ( london ) and worked with patrick blackett on nuclear physics . in 1938 after spending several years on the south sea island of tahiti , he returned to birkbeck and began working on beta - spectra . here
he developed a keen interest in electron optics . at the time of the publication of their paper reporting what has become known as the ab effect
, he was working in edinburgh university .
the collaboration between ehrenberg and siday started in 1933 when both of them were first at birkbeck , although from reading the personal recollections of werner ehrenberg , it seems that many of these discussions took place in the local pubs !
these early discussions were eventually interrupted by ray siday s adventures in tahiti and then , of course , by the war , so , in effect , they did not start collaborating again until after the war when they both returned once again to birkbeck .
their discussions were very wide ranging , but it was the principles of electron optics that was the focus of their attention and it was these discussions that led them eventually to predict what has become known as the ab effect .
their paper reporting this effect @xcite appeared ten years before the classic paper of aharonov and bohm @xcite .
i say at least , because in 1946 , siday took a three year ici fellowship at edinburgh to continue his work on beta - spectra with norman feather who , in turn , had previously worked with ernest rutherford in the cavendish at cambridge .
siday s work on beta spectra involved , among other things , the focussing of the beta rays in magnetic fields .
this is what sustained his interests in electron optics in general . in the pioneering days of designing electromagnetic lenses ,
much use was made of the analogy with optical lens systems .
of course in optics , one had long been aware of the tensions between ray optics on the one hand and wave optics on the other . in the case of electrons
there was a similar tension , between the particle properties , rays , and their wave properties . in geometric optics
the equation of the ray between two points @xmath27 and @xmath28 is obtained from the variation principle , @xmath29 this is essentially fermat s principle which is analogous to hamilton s principle in dynamics @xmath30 here @xmath1 is the conjugate momentum obtained from the lagrangian of the electron in an electromagnetic field .
ehrenberg and siday showed , in a manner that would be regarded as rather clumsy today , that this momentum would be given by @xmath31\end{aligned}\ ] ] thus they concluded that one could think of a refractive index for the electron lens as given by @xmath32 \label{e : ref}\end{aligned}\ ] ] here @xmath33 is simply the optical path length of the ray , so that if we divide this by the de broglie wave length of the electrons , we can find the phase difference between the points @xmath27 and @xmath28 on the ray .
the practical problem that siday was thinking about was the design of a magnetic lens for his beta spectrometer , so the electrostatic potential @xmath34 was put to zero .
thus one considered an optical path defined by @xmath35 $ ] .
now one can clearly see a problem .
this optical path depends upon the vector potential @xmath36 , so clearly the refractive index is not a gauge invariant expression .
this problem was clearly recognised by ehrenberg and siday and is discussed carefully in their paper . in order to motivate the discussion of gauge invariance , ehrenberg and siday recalled the mathematical conditions that must be placed on an optical refractive index .
this index , @xmath37 , is a measured quantity and so must be finite and single valued .
it must also be continuous , except at a finite number of surfaces separating any different media traversed by the ray .
the same conditions must be satisfied by the equivalent electron refractive index defined by the rhs of equation .
this means that the refractive index must be fixed everywhere in space once it is fixed in the neighbourhood of one point .
furthermore , it must be single valued .
it should have no singularities and any discontinuities should be of such a nature that they appear as limiting cases of a continuous refractive index .
since @xmath38 occurs as an additive term in the refractive index , the same conditions must be applied to it and hence to @xmath39 itself .
now these conditions are just those for the validity of stokes theorem and this is the only valid restriction which must be imposed on it .
a consequence of this is that @xmath39 can not in general be chosen so as to vanish even though the magnetic field vanishes locally .
it is this condition that is vital for understanding the ab effect .
as is well known by now , the ambiguity in @xmath39 arises because all we insist on is that @xmath40 , the magnetic field .
thus it is always possible to add to @xmath39 the gradient of a scalar , @xmath41 , without changing @xmath42 .
this follows directly from the identity @xmath43 .
this arbitrariness can not produce any observable effects in the geometrical aspects of the optics .
however wave optics is a different matter . to bring out the consequences , ehrenberg and siday consider the phase difference @xmath44 between any two paths rays .
using we have @xmath45{\bm dr}-\int _ { 0}^{1}[m{\bm v}+{\bm a}]{\bm dr'}\end{aligned}\ ] ] where @xmath46 is an element of the first path and @xmath47 is an element of the second path .
now since the momentum of the electron is constant and using stokes theorem , we find @xmath48 thus we see that , again as in the case of franz , the phase difference between the two paths depends on the flux enclosed by the closed path
. now ehrenberg and siday take the argument one stage further .
they consider the specific case when the magnetic field is confined to a region within the circuit in such a way that the electrons do not pass through the magnetic field .
in other words , if the electron should follow either path it will not experience any field .
they then ask the crucial question : can we gauge transform away the vector potential so that there is no contribution from the magnetic flux lying entirely within the circuit ? "
they show that this is not possible without introducing a singularity which would lead to the breakdown of stokes theorem .
thus they conclude that it is not possible to find a vector potential which satisfies stokes theorem and removes the anisotropy of the whole space outside the field . the fact that this irremovable anisotropy from the field free region as a whole emphasises the fact that the electron - optic refractive index contains the vector potential and not the magnetic field strength .
this led them to the conclusion that wave - optical effects will arise from an isolated magnetic field even though the rays travel in a field free region .
to emphasise this claim they sketch an experimental situation that would demonstrate this effect .
this is shown in figure [ fig : magflux ] isolated from rays , width=384 ] both ehrenberg and siday found this result very strange and totally contrary to what they would expect .
at that stage the vector potential @xmath39 was simply regarded as a mathematical symbol with no observable consequences . yet here , in quantum mechanics , it has an observable consequence .
siday was overheard by john jennings @xcite one day remarking to ehrenberg , is nt it odd that the only physical effect produced by a magnetic flux is in a superconductor ? "
ehrenberg replied ach ! but what better superconductor than free space ! "
siday went around explaining the effect to anyone who would listen and asking where the argument had gone wrong , if indeed it had .
all were puzzled and felt it was wrong , but could not put their finger on the error .
siday became incensed with the lack of enthusiasm .
there are all these sodding geniuses poncing around but if they ever have to do anything- oh yes , that s a different matter is nt it ! "
eventually he became so frustrated that he decided to present the conundrum to max born to see what he would say .
recall that at this time , siday was working at edinburgh university in norman fowler s laboratory , while born was then tait professor of natural philosophy at the same university , so he decided to invite max born to his laboratory .
david butt - rays .
butt was one of the birkbeck group that carried out one of the first experiments to test for quantum non - locality holding over distances of up to 6 m . @xcite .
] shared the same laboratory with siday and at the time of the meeting , was sitting in the corner of the lab while siday and born discussed the effect .
unfortunately he could not hear the actual conversation , which lasted about 45 minutes , but he did see born shaking his head from side to side every so often , seemingly with incomprehension . at the end of the discussion , the two of them rose , shook hands and born departed with a face looking like thunder .
as soon as the door closed , siday came up to butt and exclaimed well , sod it , i got absolutely nothing out of him . who can i ask now ? "
butt has often puzzled about born s reaction .
the problem was simple enough to describe .
all the others , mainly experimental colleagues , had understood what siday was claiming , but could not spot the ` error . '
if what siday had presented to born was an obvious consequence of standard quantum mechanics , surely born would have said so , but he did nt . nor did he offer any criticism of the explanation of the effect . in spite of this apparent indifference , ehrenberg and siday decided to go ahead and publish the work in the proceedings of the physical society , which was one of the main british journals at the time .
unfortunately they clearly had not been advised by a publicity agent as the paper was entitled the refractive index in electron optics and the principles of dynamics " , a title which gives no clue as to the radical nature of what they had discovered .
even the abstract did not highlight the effect they had found , but the conclusion was quite clear .
one should observe a fringe shift proportional to the magnetic flux included and isolated from the passing electrons .
david butt also informed me that a few years later in about 1957 , he had been talking with siday , and asked him if either he or ehrenberg had been contacted by anyone concerning the paper .
siday s is reported as saying , no .
we have not heard a bloody thing not as much as a whisper .
it has fallen to the bottom like a lump of lead " .
but why should they ? at that stage , the vector potential @xmath36 was still regarded as merely a mathematical convenience and could be gauge transformed away .
therefore it should produce no physical effect .
furthermore the effect was presented in a context that it appeared to be a problem in designing electron lenses , not a general new effect .
the choice of title of their paper only confirmed this .
however it was not only the title of the paper , the presentation suffered from two further disadvantages .
firstly the effect was discussed in specialist terms of ` equivalent refractive indices ' , using the optical path analogy for the electrons .
this was the language in common use amongst electron lens specialists at the time , but this terminology was not in general use by those working in quantum mechanics , so it gave the impression , wrongly , that it was a particular effect that was of interest only to specialists in that field .
nevertheless the fringe shift was calculated correctly and paper gave a clear discussion of the consequences .
ehrenberg and siday @xcite concluded that _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ one might therefore expect wave - optical phenomena to arise which are due to the presence of a magnetic field but not due to the magnetic field itself , i.e. , which arise whilst the rays are in field - free regions of space .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the second disadvantage was that the journal in which they published was not one of the leading journals at that time , being a publication of the british physical society before being taken over by the british institute of physics . as a consequence , ehrenberg and siday have not got the recognition that they deserve from the physics community , particularly in america . in saying this , i want to make it absolutely clear that no blame can be attached to aharonov or bohm .
bohm was unaware of the original es paper @xcite when they wrote their first paper @xcite .
they came to the same conclusion independently . in their second paper
@xcite they acknowledged the ehrenberg and siday paper had obtained the same results using a ` semi - classical ' treatment , a rather unfortunate choice of words .
the case of ehrenberg and siday falls neatly into what kossel and mllenstedt call the ` amerigo - effekt ' @xcite .
unfortunately such cases can and do happen when one is exploring a new and unexpected effect , long before the foundations of phenomenon has been properly laid .
in fact if their discussion was put into modern terms , we would see that ehrenberg and siday were exploring the common mathematical background shared by both optics and electron optics , namely , the symplectic group and its double cover , the metaplectic group @xcite .
the discussion of rays follows directly from the symplectic group .
in fact the rays are simply generated by a symplectomorphisms .
on the other hand , the wave properties follow from the covering group , namely , the metaplectic group .
what ehrenberg and siday had discovered in their own way was that the homotopy group of the covering space was non - trivial and were on the way to discovering the notion of a winding number .
alas being experimentalists , they would not have known about these advanced mathematical structures , then or even later when these techniques became more well known .
ehrenberg and siday s work remained unknown for ten years before the effect was rediscovered by aharonov and bohm @xcite .
their paper goes straight to the heart of the problem .
they note that although in classical physics the fundamental equations can always be written in terms of fields , in quantum mechanics the potentials can not be removed from these fundamental equations , therefore this must have observational consequences .
they then ask , ` what are these consequences ? '
lev vaidman , a long time collaborator of aharonov , told me that yakir had spotted the vector potential producing observable effects but , did not realise that potentials were universally considered as mere mathematical artefacts . ah !
the innocence of young research students ! he went to talk with bohm , his then supervisor , and they discussed the idea .
this discussion led them to propose an actual experiment based , in actual fact , on figure 1 that appeared in the ehrenberg and siday paper .
their proposal was to place a shield to the right of the two slits and then to place immediately to the right of the shield , in its geometric shadow , a small long solenoid with its axis parallel to the slits . to ensure none of the field of
the solenoid could spill out into the region of the electron paths , one could place a strip of mu - metal suitably shaped to conduct the field produced by the ends of the solenoid around the electron paths .
this ensures that the electrons move in a field free region .
this was precisely what was done later in the beautiful experiment carried by mllenstedt and bayh @xcite and bayh @xcite .
aharonov and bohm were rather fortunate in that an experimentalist , robert chambers at bristol university where they were working , immediately set about doing an experiment to show the effect existed @xcite .
he used a magnetic whisker and clearly demonstrated the effect . however because of the unexpected nature of the effect , people argued that as the magnetic whisker produced an unshielded field , the effect may be due , after all , to the field rather than the potential .
this was wishful thinking .
however the appearance of bayh s results immediately showed that any arguments about stray fields causing the effect could be ruled out . since those early days a number of more refined experiments
have all confirmed the effect .
the full details of all these experiments can be found in a review article by olariu and popescu @xcite .
it is now clear that the ab effect arises directly from the schrdinger equation as was explained in the first aharonov and bohm paper @xcite . yet clear as the paper was , it too was received with some scepticism and even opposition to begin with .
for example , victor weisskopf @xcite wrote in some brandies lecture notes , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the first reaction to this work is that it is wrong ; the second is that it is obvious . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this effect is now considered to be of great theoretic importance as it is the first example of quantum gauge phenomena .
gauge theories have become central to the modern theory of particle interactions , spawning many examples of gauge phenomena @xcite .
the importance of the effect is reflected in a leader article published in nature where it was proposed that aharonov and bohm should share the nobel prize with michael berry for their contributions to the understanding of gauge effects .
i asked bohm for his reaction to this suggestion .
he replied that he did not think the ab effect alone was that noteworthy and added : after all it is only a straight forward application of standard quantum mechanics , and anyway ehrenberg and siday were there first ! " for me that sentence the way physics should be done with , humility , generosity and honesty .
i should like to thank my colleague david butt without whose detailed reminiscences of his interaction with ray siday , without this background , this paper would have been a poorer report .
i am also indebted to the late john jennings for his input and translation of the abstract of franz s talk given in danzig in 1939 .
bayh , w. , messung der kontinuierlichen phasenschiebung von elektronenweuen i m kraftfeldfreien raum dutch das magnetische vektorpotential einer wolfram - wendel , _ zeit .
fr physik .
_ , * 169 * ( 1962 ) 492 - 510 .
franz , w. , electroneninterferenzen i m magnetfeld , _ verh .
d. deutschen physikalischen gesellschaft _ * 20 * , ( 1939 ) 65 - 6 . and
_ physikalische berichte _ , 21st annual volume ( 1940 )
these references are identical in content .
mllenstedt , g. and bayh , w. , messung der kontinuierlichen phasensehiebung von elekfronenwellen i m kraftfeldfreien raum durch das magnetische vektorpotential einer luftspule _
naturwissenschaften _ ,
* 49 * ( 1962 ) 81 - 2 . | this paper traces the early history of the aharonov - bohm effect .
it appears to have been ` discovered ' at least three times to my knowledge before the defining paper of aharonov and bohm appeared in 1959 .
the first hint of the effect appears in germany in 1939 , immediately disappearing from sight in those troubled times .
it reappeared in a paper in 1949 , ten years before the defining paper appeared . here
i report the background to the early evolution of this effect , presenting first hand unpublished accounts reported to me by colleagues at birkbeck college in the university of london . |
the relativistic heavy ion collider was completed in 1999 and a host of new experimental results soon appeared . during the past year , the nuclear physics community has undertaken the task of formulating a long range plan to guide funding priorities for the next 10 years .
what follows is my view of the future .
i have chosen to be rather speculative at times .
i will first review what we have learned so far , then pose the questions formulated in the long range plan ( lrp ) , and use these to guide a discussion of the future measurements which should be made .
in addition to the many recent experimental results , theory has been making rapid progress in explaining the phenomena we are seeing at rhic .
this is possible for two reasons .
first , powerful theoretical techniques , have been brought to bear on the problem .
these are reviewed elsewhere in these proceedings by a dimitru@xcite .
secondly , rhic has taken us into a regime in which these theoretical techniques are applicable .
it is the convergence of theory and experiment which leads to new understanding .
the collection of experimental data alone makes us simple accountants of miscellaneous facts . on the other - hand
a non - verifiable theory may be elegant , but will not give us an understanding of the world in which we live .
i would like to suggest that a sort of standard model is just beginning to emerge in our understanding of the evolution of relativistic heavy ion collisions in the regime we have entered .
this picture is not entirely new .
the general picture of the phase diagram of qcd has been around for many years .
what is new , is that rhic , now takes us into a regime in which theoretical techniques can now be used to do reasonably reliable calculations , using the model , of experimentally measurable quantities .
as such we are beginning to enter into an era of precision measurements in which theoretical predictions of a fundamental nature can be made , and experiments can be done under controlled conditions so that these ideas can be tested .
to do many of these measurements , rhic will need an increase in luminosity .
what is being proposed by the community is an increase by a factor of 40 by the use of electron cooling .
in addition , experiments will need to be upgraded , to enhance capabilities and increase sensitivity for rare or hard to measure probes , and to increase the ability of detectors to handle a higher data rate through the use of higher level triggers and expanded daq .
the details of this will be covered in the next talk by t. ludlam@xcite .
i will first summarize of some of the results from rhic as they stand now .
much of this material is covered in more depth in separate contributions to this volume . unless specified , the data is from the first run where the center - of - mass energy was 130 gev .
a small amount of data is now available from run-2 where the center of mass energy was at the canonical value of 200 gev .
needless to say , the various rhic collaboration are working hard to produce the results from this latest run .
the first task is to make an experimental estimate of the initial energy density of the system formed in collisions at rhic to see whether it is high enough to induce a phase transition .
lattice calculations indicate that the phase transition should occur between 150 and 200 mev corresponding to energy densities between 0.6 and 1.8 gev/@xmath0 with the lower estimate being favored by models which include light fermions .
bjorken has given an estimate of the energy density assuming boost invariance@xcite @xmath1 where r is the radius of the nucleus in question , and @xmath2 is the time after the initial collision of thermalization when a temperature can first be defined , typically taken to be between 0.2 and 1 fm / c . as theory
has improved it has been recognized that there are at least two relevant time scales for @xmath2 .
the first , as described below is a formation time related to the colored glass condensate .
the second , is the actual thermalization time .
the phenix collaboration has measured @xmath3 , giving an energy density times thermalization time to be @xmath4 .
conservative estimates give @xmath5 , whereas estimates from the equilibration of gluons give @xmath6 .
this gives an energy density estimate of between 5 and 23 gev/@xmath0 , an order of magnitude greater than that required for the phase transition as predicted on the lattice .
recently , mclerran@xcite and his colleagues have proposed that the initial conditions in heavy ion conditions could be calculated assuming that the gluon distributions at low x are saturated .
the gluon distributions at low - x for protons @xmath7 , hence violate unitarity at very low x and saturate . in protons
it will occur at about @xmath8 depending on the q@xmath9 . in a lorentz contracted nucleus
, there is essentially a geometric magnification factor of a@xmath10 making the relevant @xmath11 .
mclerran and his colleagues assumed that since the occupation numbers were high , one could use a classical approximation .
this leads to non - perturbative calculations of various experimentally measured quantities , using renormalization group methods which depend on a scale q@xmath12 , the saturation momentum given by @xmath13 .
at rhic , this is around 1 - 2 gev / c .
they have dubbed this state , a colored - glass - condensate for reasons i will not explain here .
one of the first calculations to be done using these methods was for the multiplicity distributions as a function of centrality@xcite at @xmath14130 gev ( figure [ fig1]a ) .
admittedly this was a post - diction after the data was available .
however , kharzeev and levin@xcite then predicted the centrality dependence at the higher energy of @xmath14200 gev .
the prediction , shown in the right panel of the figure does extremely well for the recently published central data from the phobos collaboration
. it will be important to observe whether the centrality dependence at the higher energy is borne out .
the model gives us an initial energy density of 18 gev/@xmath0 in good agreement with the estimate above .
this corresponds to a formation time of @xmath15 .
presumably equilibration happens somewhat later .
this time corresponds to the time in which the gluons `` go on mass shell '' , hence this energy density refers not to a thermalized energy density , but rather the maximum energy density reached by the system .
kharzeev , levin , and nardi@xcite have also made a prediction at @xmath1422 gev , but they point out that the model may not work since @xmath16 becomes large .
their central prediction at this energy is that the number of charged particles per participant at mid - rapidity rises slowly by about 20% for central collisions due to the running of the coupling constant @xmath16 .
it is worth examining the source of the centrality dependence of the multiplicity in the cgc model . in this model
the number of gluons per participant is proportional to the inverse of the strong coupling constant hence , somewhat schematically , @xmath17 .
the value of @xmath18 rises as a function of centrality since the quantity of interest is the number of gluons per unit area in the target nucleus as seen by a particular projectile .
the more the number of gluons , the higher the saturation scale .
this is shown in figure [ fig11 ] where @xmath19 is plotted versus n@xmath20 .
the corresponding value of @xmath21 is also plotted and compared to the multiplicity per participant as measured by the phobos collaboration .
the shapes of the distributions are similar since @xmath22 calculated in the cgc model is proportional to @xmath21 .
this gives rise to the agreement between the theory and the data shown in figure [ fig1]a .
this result connects the centrality dependence of the the multiplicity per participant to the running of the coupling constant .
if @xmath16 did not run , there would be no centrality dependence .
it is interesting that a bulk property such as the total multiplicity is intimately connected with a fundamental property of qcd - the running of the coupling constant .
the calculation relies on the fact that the number of produced particles is essentially equal to the number of gluons liberated in the initial collision , which in turn is equal ( actually a factor of 2ln2=1.34@xcite ) to the number of gluons in the initial virtual saturated state .
this then implies that the number of particles may be conserved through the parton to hadron transition - a miraculous fact noted in reference @xcite .
one can speculate why this is so .
an understanding of this may lead to insights into the chiral and deconfinement phase transitions , since the process of partons turning into hadrons includes the generation of the dressed masses of the quarks which comprise hadrons and the confinement of quarks into hadrons .
such a conservation may not hold for other quantities such as the momentum distributions of the final state particles which may be influenced by later stages of the system .
the colored glass condensate model nicely explains one of the seemingly disappointing early results from rhic .
the charged particle multiplicity was found to be lower than naive expectations which sprung from entropy arguments invoking the fact there would be more degrees of freedom in a system of quarks and gluons than in a system of hadrons . in the cgc scenario , the production of entropy is simply choked off by the saturation .
early results from both the star and phobos collaborations indicated that there was a surprisingly strong elliptic flow in non - central collisions ( see figure [ fig2 ] ) .
elliptic flow is an anisotropy which exhibits itself in the momentum distribution of the produced particles .
this anisotropy in momentum reflects a space anisotropy of the original , almond shaped collision region . for the space anisotropy to be transfered to the momentum , the system must equilibrated .
since flow is a collective effect , and each particle s direction is defined by the spacial distribution of all the particles , each particle must `` know '' the shape of the original distribution .
the effect is self quenching .
if the time for equilibration is too long , the collision region looses its almond shape , hence the momentum distribution looses its anisotropy .
a strong elliptic flow , therefore , implies an early thermalization .
the details of this argument were developed in a more rigorous fashion by u. heintz and his colleagues@xcite .
one can see in figure [ fig2]a , that in mid - central collisions , the data gives a value that is consistent with hydrodynamics which of course assumes thermalization with a zero mean free path .
this hydrodynamical model@xcite requires a thermalization time of 0.6 fm / c giving an energy density of 20 gev/@xmath0 , again consistent with the previous estimate .
it seems fairly certain , then that the initial state formed in au - au collisions at rhic is a strongly interacting system at very high energy density .
we can begin to assemble a model ( figure [ fig3 ] ) of rhic collisions beginning as a colored glass condensate - essentially a saturated condensate of gluons .
this is a virtual state since it is presumably there in all nuclei .
what is important in collisions at rhic , is that the colliding system provides a mechanism for the virtual gluons to become `` real '' in a formation time which is of order 0.1 - 0.2 fm / c .
these `` real '' gluons , then form the initial state which is of interest to us . at this point
they are not equilibrated - they have just come into existence at the beginning of the second stage .
the mechanism and timescale of thermalization is probably the most critical question to answer in the short term .
if thermalization of gluons occurs then we have the inevitable conclusion that we have created a quark - gluon plasma .
the parton cascade model@xcite showed the importance of radiative processes in equilibration .
recently , mueller and his colleagues@xcite have shown this by analytical techniques as well .
they give an estimate for the equilibration time as 1/q@xmath12 multiplied by some power of @xmath16 .
the monte - carlo implementation of the parton cascade model gives an equilibration time of @xmath23 0.3 fm / c , for a total formation time plus thermalization time of about 0.5 -0.6 fm / c .
this is consistent with the value found from the elliptic flow results and hydrodynamic models of 0.6 fm / c , however this agreement should probably not be taken too seriously until more reliable estimates for the thermalization time can be made .
we are left then , at a time of about 0.6 fm / c with a plasma of primarily gluons .
presumably quarks begin to appear as the chemical reactions continue .
one of the interesting characteristics of this state is that value of the quark condensate @xmath24 is zero and chiral symmetry is restored .
the question of how the condensate actually develops is an interesting one which we know very little about . the equilibrated system , which we can now characterize as a qgp expands and generates flow as it cools .
the system proceeds through a mixed phase as it begins to hadronize .
this is arguably the most interesting time in the history of the collision - the third and fourth stages in figure [ fig3 ] . at this point
we are faced with host of questions .
is the transition first or second order , or simply a cross over ?
we do not know the relationship between the deconfinement transition , responsible for the binding of quarks into hadrons , and the chiral transition , responsible for the generation of hadronic mass .
are the transitions at the same value of the temperature ?
what is the latent heat ?
is there an inflation related to the phase transition as is presumably true in the early universe ?
are there large scale fluctuations ? answering these questions are the primary goals for the present and future program at rhic .
the system continues to cool and expand in the hadronic phase until eventually all the interactions cease and the particles freeze - out .
it is these hadrons which we can actually detect .
they should reflect the state of the system at hadronic freeze - out .
rhic has made impressive progress in the measurements of these hadronic observables .
more complete descriptions are elsewhere in these proceedings .
i will only give a brief overview .
figure [ fig4 ] shows the plethora of particle ratios that have already been measured .
these values are compared to a model assuming chemical equilibrium with three parameters : the freeze - out temperature t , the baryo - chemical potential @xmath25 , and a strangeness saturation parameter , @xmath26 .
first , one observes that the model works reasonably well and unlike at the sps , strangeness is saturated , i.e. @xmath27 .
the fit gives a temperature of 190 mev and a baryo - chemical potential of 0.04 at chemical freeze - out .
this is consistent with a near baryon free system - a fact which can be deduced from the value of @xmath28/p of 0.6 measured by several of the rhic collaborations .
we can look at thermal equilibrium as well .
figure [ fig5]a shows a fit to the inverse slopes , for central collisions in a model assuming a single freeze - out temperature and radial flow for a variety of center of mass energies from gsi to rhic .
one observes that the temperature appears to saturate at a value of about 130 mev at the ags and remains the same at the sps and rhic . at rhic however , the value of @xmath29 is noticeably larger .
( i note that the error bars for @xmath30 and t@xmath31 are correlated .
if one held t@xmath31 constant the errors on the flow velocities would be considerably smaller .
this can be seen by simply examining the original inverse slopes . )
it appears that at rhic we are observing an increasing radial expansion .
a intriguing speculation is that this expansion might be the result of `` inflation '' caused by the chiral phase transition , similar to that which caused the inflationary phase of the early universe .
one of the questions that one might ask is whether this rapid radial expansion develops in the early gluon phase of the system , during a mixed phase , or later during the hadronic phase before freeze - out .
figure [ fig5]b , shows the actual inverse slopes measured for a variety of particles as a function of the mass .
the strange - baryons show a tendency to be below the line corresponding to a particular temperature and radial flow velocity .
this might indicate that these particles were freezing out somewhat earlier then their lighter counterparts at a time when the flow had not had time to develop fully .
a similar trend was seen at the sps .
this interpretation would imply that at least some part of the radial flow developed during a hadronic stage .
let us return to the early stages of the collision .
are there probes which can provide information about the state of the system and substantiate the ideas outlined earlier , that is to study the state which may be the qgp ?
of course the answer is yes - these are the penetrating probes - leptons , photons , and most notably high p@xmath32 particles - presumably the leading particles of jets .
perhaps the most striking result from the recent data at rhic comes from the observed suppression of high @xmath33 particles , particularly @xmath34 s from the phenix collaboration .
x.n . wang and others have pointed out that high p@xmath32 jets could be used as probes of the early stages of heavy ion collisions@xcite .
high p@xmath32 partons are the result of hard collisions which happen very early in the collision .
hence as they traverse the fireball , their characteristics can give us information about the system from the earliest times .
in particular , the energy loss of these partons should be much greater in a system of deconfined quarks and gluons , since they are directly exposed to the color charges of the medium .
because the jets resulting from these high p@xmath32 partons are difficult to isolate from the soft background , it was suggested that leading particles could be used since they typically take about half the energy of the jet .
early calculations simply assumed a constant energy loss per unit length .
more detailed qcd calculations@xcite indicate the energy loss actually depends on the total path length of the parton through the fireball .
this rather non - intuitive fact is a result of interference effects between the soft gluons radiated from the parton as it looses energy .
the energy loss in a qgp can be extremely high - greater than 10 gev / fm at the highest energy densities presumed to be created in these collisions .
however , since the system expands and the lifetime of the high density stage is rather short , the average energy loss of a parton is considerably less . before looking at the data , it is worth briefly describing the assumptions used to scale pp collisions to au - au collisions .
there are two basic scaling methods .
the first , and most familiar to heavy ion physicists is scaling with n@xmath20 , appropriate for are thermal distributions coming from soft processes .
momentum distributions are usually plotted as a function of the transverse mass , @xmath35 , a practice coming from that fact that all particles regardless of mass could be described as a universal exponential in @xmath35 for lower energy pp collisions . in heavy ion collisions exponentials tended to describe the shape of the spectra , however the slopes diverge with the heavier particles having larger inverse slopes .
this was attributed to the effect of hydrodynamic flow tending to give a higher transverse momentum to heavier particles .
these processes reflect the bulk nature of heavy ion collisions .
much of the data from the ags and sps scale as a function of the number of participants , n@xmath20 , which in a central au - au collision is 2 * 197 .
a notable exception to this was the production of strange particles which scaled faster than this , and has been attributed to the enhancement of strangeness in heavy ion collisions .
the second scaling method is more appropriate for hard processes - i.e. jet production , and is of course more familiar to those working at high energies . here
spectra can be described as a power law p@xmath32 .
there are no collective effects since one assumes that there is essentially one hard interaction .
particles originate from jet fragmentation . here
the scaling to heavy ion cross sections from pp goes as the number of binary collisions - i.e as the number of hard collisions .
a central au - au collision has about 1000 binary collisions .
analysis of multiplicity distributions provide evidence for a part of the cross section scaling as the number of binary collisions@xcite .
figure [ fig6 ] shows the @xmath34 p@xmath32 distribution together with the expected value from the scaling of the pp cross section with the number of binary collisions@xcite for both peripheral collisions and central collisions .
as one can see , the the scaling works well for peripheral collisions . in the central collisions ,
however , the data is over - predicted .
charged particles show a similar behavior , with some complications , as i will describe in a moment .
the effect can be highlighted by dividing the data by the scaled pp distribution as in the right panel , shown now both for @xmath34 s and charged particles . in both cases
the result is considerably below one .
also shown is the range of results compiled from data at the sps .
the data from the sps below @xmath36 also lies below one , since the production mechanism for these lower energy particles scales as n@xmath20 .
these are the vast majority of particles .
we are concerned with the small number of high p@xmath32 particles above @xmath37 .
( the exact value of `` high '' p@xmath32 is open to question ) . at the sps what we see
is an enhancement of high @xmath33 particles coming from quark scattering - the well known `` cronin '' effect .
what is seen at rhic is quite the opposite .
there is a strong suppression above 1.5 gev for both the charged and neutral particles .
such a suppression may be the first indication of a large energy loss in a qgp .
one may then inquire as to the difference between the charged and neutral spectra .
figure [ fig7 ] shows the charged spectra for minimum bias events from phenix , now using particle i d to the extent possible .
one can immediately see the problem . at momenta above 2 gev
, protons and anti - protons dominate the spectra
! how can this happen ?
this certainly is not characteristic of the distributions of leading particles in jets , where only 10% is a baryon@xcite .
it may be that collective effects such as radial flow are actually pushing the baryons to a higher p@xmath32 and we are comparing particles which come from different sources - i.e. jet fragmentation after energy loss for the pions , and low energy protons and anti - protons which have been accelerated due to radial flow effects . at the same time
the high p@xmath32 partons themselves loose energy and their daughter pions will have a more thermal spectrum then would be characteristic of jet fragmentation .
one of the unanticipated implications for the future , is that particle identification at high @xmath33 will be critical to the upgrades considered at rhic .
data on leptons and photons will be forthcoming .
already phenix has published the first results of electron production giving limits on open charm .
this is of particular interest because of the enhancement of intermediate mass dileptons at cern . however , since charmed quarks themselves pass through the medium just as any other quark , they should loose energy as well and be suppressed ; though this effect could be minimized due to a `` dead cone '' effect@xcite . within the next year
we should begin to see results on j/@xmath38 production in both the electron and muon channels , low mass vector mesons to the di - electron channel , and perhaps thermal dileptons and direct photons .
if the view espoused here of the high p@xmath32 spectra prevails , it will provide vindication of the model described previously and lead to more conclusive results regarding the possible formation of a quark gluon plasma .
perhaps , as i have indicated , it raises more questions than it answers . as a part of the task of putting together the long range plan for nuclear physics
several meetings were held to consider the future of of the rhic program . in doing so , the following four major questions were formulated .
* _ evolution of the system : _ in relativistic heavy - ion collisions , how do the created systems evolve ? does the matter approach thermal equilibrium
? what are the initial temperatures achieved ? * _ deconfinement : _ can signatures of the deconfinement phase transition be located as the hot matter produced in relativistic heavy - ion collisions cools ? what is the origin of confinement ? * _ chiral symmetry restoration : _ what are the properties of the qcd vacuum and what are its connections to the masses of the hadrons ? what is the origin of chiral symmetry breaking ? * _ matter at the highest energy densities : _ what are the properties of matter at the highest energy densities ? is the basic idea that this is best described using fundamental quarks and gluons correct ? in what remains of this paper ,
i would like to take each of these questions in turn and make some suggestions about measurements which might help us answer these questions , and point out what upgrades will be necessary . in any set of such suggestions , there is a danger of compartmentalizing certain results to answering certain questions . in reality
our subject is rich with overlapping subjects - measurements meant to answer questions about the later hadronic part of the collision may lead to information about the early stages of the collision .
i will also give more detailed examples of what might be done in three particular measurements - low mass vector mesons , onium suppression , and jet suppression - keeping in mind the statements i have made about precision measurements .
i will conclude with a suggested set of priorities for our community .
this is perhaps the question for which the suite of detectors at rhic are most capable of answering , at least for the later hadronic stage of the collision .
we already have a basic understanding of much about the final freeze - out phase .
it appears that the final state hadrons reflect a both chemical and kinetic thermal distribution .
we have first estimates of the kinetic and chemical freeze - out temperatures , chemical potentials , and expansion velocities .
there is much left to understand , however , we have the tools in hand to do much of the work .
detectors are well equipped to study hbt , flow and particle yields and spectra .
thanks to improvements in theory , we may be close to some understanding of the initial stage of the collision as well . detailed tests of the cgc model will be made in future ea machine proposed to be done at rhic and in pa collisions with upgraded detectors for looking at collisions involving small x partons .
what is missing is a clear understanding of the stages , after the formation of the cgc particularly related to the thermalization of the initial gluon distribution .
does the strength of the elliptic flow signal really imply that thermalization occurs as early as 0.6 fm ?
if so , it may be that we already have the answer in hand but it is important to gather more information .
one of the possibilities for examining this stage of the collision is though real and virtual photons from quark - anti - quark annihilation , materializing as electron or muon pairs radiating from the hot , dense qcd matter .
while such radiation is emitted at all times during the collision , the reaction dynamics favors emission from the hottest part of the colliding system .
thus , measurement of the distribution of the blackbody thermal radiation will give us information on the initial temperature , and possibly the black body nature of the spectrum .
if so , these photons would play the same role for us that the cosmic microwave background radiation did for our understanding of the early universe .
the background to such a signal is formidable since photons and electrons are copiously produced from other sources such as @xmath34 decay .
upgraded detectors designed to reject such backgrounds will be necessary .
systematic analysis , and variation of the initial conditions will be required to solidify the interpretation . as mentioned previously
, the suppression of high p@xmath32 hadrons from the energy loss of partons as they pass through a deconfined medium is an important signal of deconfinement .
this may be enough to establish the effect , however it will be important to be more quantitative .
this can be done by making measurements of direct photons produced opposite high transverse momentum hadrons . since the photon recoils against the quark jet , and
since it does not suffer energy loss in the deconfined medium , the photon serves as a indicator of the initial transverse momentum of the jet .
this will provide a means to make careful , quantitative measurements of the energy loss .
one interesting possibility is to flavor tag the high transverse momentum hadron . a leading k@xmath39 with no valence quarks is more likely to come from a gluon jet .
this would allow one to measure the difference in the energy loss between gluon and quark jets .
gluon jets are expected to loose energy at twice the rate of quark jets in a deconfined medium .
these measurements will require high luminosity because of the low cross sections of such high p@xmath32 events .
in addition detectors will be required to handle the large luminosities , as well as to identify hadrons to high p@xmath32 .
j/@xmath38 suppression is another well - known signature of deconfinement .
phenix will be able to measure j/@xmath38 production in both the muon and electron channels .
star will have access to the electron channel within the next several years as their electromagnetic calorimeter is completed providing a second measurement of this signature .
one of the critical measurements that must accompany the measurement of the j/@xmath38 is that of open charm production to see whether open charm is enhanced or comes from the `` melting '' of the j/@xmath38 . to do this
, specialized vertex detectors must be added with the position resolution that would allow a measurement of the charm vertex separated from the original event vertex .
star , phobos and phenix have all embarked on r and d programs to construct such an upgrade .
the j/@xmath38 is but one of the vector mesons in the charm family .
the excited states of the j/@xmath38 as well as the @xmath40 family will all exhibit some degree of suppression . the suppression of the associated @xmath41 and @xmath42 states , can also be observed since they decay to the detectable vector mesons . each of these states will `` melt '' at a different temperature ( figure [ fig10 ] ) .
in fact the @xmath40 will be used as a control since it should not be suppressed at all at rhic energies . by varying the temperature of the system though changes in beam energy and species
, one can change the pattern of suppression of the various states . not only would this be a convincing signature of a phase transition
, it would give a good measure of the actual energy density .
it appears that the densities are high enough , even in mid - central collisions where the reaction plane can be obtained , that there is a qgp formed albeit with a smaller volume then in very central collisions .
this then raises the possibility that one can actually do tomography on the collision region and map out its density profile to some degree in the following manner .
one would measure the angle between the reaction plane and the probe of interest , for instance a high pt particle opposite a photon and measure the energy loss as a function of the reaction plane .
the path length perpendicular to the reaction plane is longer than in the direction of the reaction plane because of the almond shape of the collision region in semi - central collisions . as an example of a particularly interesting question that this could address
is to see the relationship between energy loss and path length on the one hand and energy density on the other by examining the differences in energy loss as a function of the path length .
this would enable one to check the path length squared dependence of the energy loss , mentioned before .
the measurements mentioned in this section would require measuring small cross section processes in many bins of centrality , p@xmath32 and reaction plane .
this demands an increase in the luminosity as well as large acceptance detectors capable of measuring leptons , photons , high pt particles with particle identification .
chiral symmetry is broken through the creation of a vacuum scalar condensate that couples to hadrons and provides most of their mass .
the challenge for rhic experiments is to search for evidence of in - medium mass changes of the low mass vector mesons associated with the restoration of chiral symmetry .
a direct measurement of the mass of light vector mesons such as the @xmath43 , @xmath44 , and @xmath45 is possible since they decay rather rapidly within the fireball created at the time of collisions and before hadronization .
the decay to di - electrons is particularly interesting since electrons should not be re - scattered in the medium and their invariant mass should reflect the mass of the vector meson in the altered vacuum state . since some fraction of the vector mesons decay outside the medium ( in the case of the @xmath44 some 70 - 80% ) , these can be used as a calibration point for the measurement .
the fraction exhibiting a shifted mass should change as a function of the transverse momentum and the size of the central fireball .
this would be a particularly dramatic signature of the altered vacuum .
as in the case of the thermal di - electron signal , a major upgrade will be needed to reject background for detection of the @xmath43 , the shortest lived , and hence the broadest of the vector mesons unless it is strongly enhanced which it may well be .
the presence of a phase transition as the system cools is also expected to cause fluctuations , which may survive the hadronic phase as fluctuations in particle number and type .
fluctuations and droplet formation are of particular interest , since similar processes may account for much of the large scale structure of the universe and the inhomogeneities observed in the cosmic microwave background . a variety of fluctuations have been proposed as a signature of a phase transition . if the transition is first order , the growth of hadronic droplets and the shrinking of quark - gluon droplets may yield a lumpy final state and large fluctuations in particle number .
if the transition is a smooth but sufficiently rapid crossover , domains of misaligned chiral condensate may be formed .
if the transition occurs near a critical point separating first order behavior from crossover behavior , long wavelength fluctuations imprint unique signatures on the momenta of soft pions .
experiments will search for such phenomena , and correlate their appearance with other quark - gluon plasma signatures .
the models of chiral symmetry restoration have followed essentially two approaches .
the first , typified by rapp and wambach s model@xcite assumes hadronic degrees of the freedom as the staring point .
the other , typified by brown - rho scaling@xcite , uses primarily quark degrees of freedom . under the hadronic description
, the restoration of chiral symmetry means that the vector ( e.g. the @xmath43 ) and the axial vector ( e.g. the @xmath46 ) will become degenerate when chiral symmetry is restored .
typically the particle widths increase and the mass does not go to zero .
brown and rho however , assumed that all particle masses scale with the chiral condensate which of course goes to zero when chiral symmetry is restored .
these two views are of course , seemingly contradictory - a situation that is somewhat endemic to this subject .
the two groups have made the conjecture that there is a quark - hadron duality and that the two approaches will give the same results if one did not have to resort to perturbation theory , but somehow could calculate the results directly which , of course is probably not possible , at least not analytically .
the only difference between the two approaches , then is simply the difference between the sets of basis states or `` degrees of freedom '' that they choose to use to solve the problem .
let us suppose that the chiral and deconfinement transition occur at the same place .
let us also assume that the rapp wambach model was correct below the phase transition , where at least at very low temperature their set of basis states seem reasonable , and brown rho scaling is appropriate above the transition in the sense the the quark masses scale with the chiral condensate .
the rapp wambach model predicts that as we approach the phase transition from below the vector and axial vector particles widen and become degenerate . above the phase transition
, the quarks deconfine and presumably the light quark masses will go to their bare quark masses - essentially zero on the scales which we are considering . at the phase transition ,
somehow these pictures must merge .
it may be that neither set of basis states are the correct ones to use near the phase transition . if one was able to find the correct basis states , one may actually be able to do the calculation in the region of interest .
a similar situation was faced by condensed matter theorists in trying to explain super - conductivity .
bardeen , cooper and schrieffer managed to guess the correct set of basis states to use - cooper pairs .
presumably they could have constructed their theory using electrons as the basis states .
the theory would have been just as good but it would have been hopelessly complicated .
finding the right set of basis states is one of the major steps to solving such a problem . in the case of qcd
, it may be that neither the hadronic nor the quark basis are the best set of basis states to use near the phase transition .
it also may not be possible to guess the right set of such states with knowledge available to theorists now .
what is needed are experimental results , which then can used as a foundation for making a reasonably educated guess .
of course we can approach this problem in a brute force manner using lattice gauge theory on massive computers , however finding the right degrees of freedom or basis states gives one a much better qualitative understanding of the phenomena .
this is one of the instances in which the interplay between theory and experiment is the most productive .
the detection of low mass vector mesons , particularly the @xmath43 , decaying to di - electrons is a challenging task for similar reasons to the the problem of identifying the thermal di - electron signal mentioned previously- the backgrounds are fierce from @xmath34 dalitz decays and photon conversions .
a scheme of rejection of dalitz and conversion pairs must be implemented .
the @xmath43 , in addition to being useful for the understanding of chiral symmetry restoration , can also be used as a clock of the mixed and hadronic phases of the collision . since
the lifetime is so short , the @xmath43 decays and is regenerated several times during the collision - for instance from @xmath47 interactions . since electrons do not interact strongly they can emerge from the collision relatively unhindered .
a long mixed and hadronic phase should enhance the @xmath43 to di - electron signal in a matter which yields the lifetime . in order to study some of these effects , and the need for dalitz rejection
, a simulation was done using as input , the model of rapp and wambach which includes the restoration of chiral symmetry and the enhancement of the @xmath43 due to the aforementioned effect .
figure [ fig8]a shows the resonance shape of the @xmath44 meson for a variety of temperatures .
one can see the slight mass shift , and the broadening of the line shape .
a run of @xmath48 central events ( about 2 years of running ) was assumed .
figure [ fig8]b shows the signal for the @xmath43 , @xmath44 and @xmath45 assuming the current phenix detector which has no dalitz rejection .
the first thing to note is that if the enhancement of the @xmath43 is as strong as the rapp - wambach model indicates , the signal will be clearly measurable even with the current configuration of the phenix detector without dalitz rejection .
to illustrate the difficulties , however , we turn to the @xmath45 which shifts in a manner similar to the @xmath44 .
figure [ fig8]c shows a comparison of the signal from low - pt central events where the effect should be the strongest , to the high - pt peripheral events where it should be essentially normal .
the statistical error bars are very poor due to the large subtraction which must be performed due to the background .
clearly , one would not be able to deduce any difference .
a 90% dalitz rejection was assumed for figure [ fig8]d , and a clear effect of width broadening due to chiral symmetry restoration can be seen .
there are essentially two schemes being considered for dalitz rejection .
one of the methods contemplated is that of a hadron blind detector with large acceptance which has a high probability of detecting both particles of a dalitz pair .
this typically involves a field free region near the vertex so that the low momentum partner is not bent out of the acceptance of the detector .
since the detector is sensitive to electrons which have a relatively low multiplicity , electrons and positrons can be reconstructed and rejected if they are consistent with a dalitz of conversion pair .
such a scheme could even be implemented on - line .
the second involves the use of a vertex detector together with the ability to do good particle identification .
such a detector could measure low momentum electrons since it is so close to the vertex .
the background pair could then be rejected in a similar manner to the hadron blind scheme .
there is a great deal to learn about hadronic interactions at very high energies where we are probing the very low x region of the hadron wave - function .
the cgc is critical to understanding these process and an understanding of the gluon densities at these high energies forms the basis for the calculations of the initial conditions in nucleus - nucleus collisions .
critical experiments to perform will include pa and ea collisions
. a future project at the rhic collider will be the addition of an electron ring to study ea collisions . in the near future we can begin to use pa collisions .
proton - nucleus collisions are necessary component of the heavy ion program as well , since pa as well as pp collisions serves as a basis for the normalization of aa collisions to provide an understanding of phenomena such as nuclear shadowing .
an increase in luminosity is necessary for this task since the same rare signals must be detected .
signals which can be used to measure the relevant gluon distributions primarily come from gluon fusion processes .
these include the production of heavy quarks and w bosons and direct photons .
anti - quark distributions can be probed via drell yan .
saturation will affect the rate and the transverse momenta of these probes .
processes which involve very low -x partons will occur in the very forward portion of the detector and upgrades to lepton , photon and micro - vertex detectors which can cover pseudorapidities as high as @xmath49 are required .
we are now left with the task of setting the priorities for the future of the rhic program .
one of the important aspects providing enormous strength to the program is a redundancy in the capabilities of the detectors .
the different detection and analysis techniques as well as different regions of acceptance can provide a cross check and corroboration of critical signatures lending credibility to the discovery of new phenomena .
in particular , star is now adding electromagnetic calorimetry over a large acceptance , providing balance to phenix s aggressive electron and muon identification over a somewhat smaller region .
r and d for the detector upgrades has begun .
funding must be provided for this .
upgrades which will be added should and does include a program for dalitz rejection , high p@xmath32 particle identification , and micro - vertex detection for heavy quark production .
weaknesses of the various detectors are being addressed - for instance in the addition of electromagnetic calorimetry for star as just mentioned , and the addition of 4@xmath50 tracking for the phenix detector to enhance jet detection capabilities .
star already has a good acceptance at very forward rapidities .
phenix is contemplating the addition of very forward detection of muons and photons as well as a micro - vertex detector in the forward region , though these projects are not as well formulated as some of the others.all detectors must of course enhance their data - acquisition capabilities to enable them to take advantage of the planned luminosity upgrade .
rhic provides us with a unique capability to understand qcd interactions , both at high temperatures and in bulk , which was the primary focus of this paper .
it will also study pp , pa , and ea collisions in which we can study the quark and gluon distributions themselves as well as the spin of the nucleon in the very rich polarized proton program which i have not mentioned at all .
i remain optimistic about many of the ideas as explained in this paper , particularly the colored glass condensate .
however it must be emphasized that it is far too early to assume that it is a well established theory .
only further experimental evidence can tell us whether it is right or wrong .
in fact it may well be that the cgc model turns out to be incorrect , however , the important point is that the future of the rhic experimental program ( and lhc ) is very bright .
the future of theory in the field is progressing rapidly as well and hand in hand , theory and experiment will work together such that perhaps in the next 5 - 10 years , our picture of high - temperature qcd will be much more complete giving us an understanding of the nature of the phase transitions and the nature of confinement and hadronic mass - which i remind the reader makes us most of the ordinary matter we see around us .
i would like to thank the numerous theorists who have patiently explained much of this material to me and corrected many of my misconceptions .
i would like to particularly thank larry mclerran who read much of the introductory material and reigned is some of my enthusiasm for the absolute reliability of qcd calculations for the energies available at rhic and reminded me that our task was not so much to create or `` discover '' the quark - gluon - plasma but to understand it .
any errors remaining in the text are mine .
l. mclerran , r. venugopalan , _ phys .
rev . _ * d49 * ( 1994 ) 2233 ; l. mclerran , r. venugopalan , _ phys . rev . _
* d49 * ( 1994 ) 3352 ; l. mclerran , r. venugopalan , _ phys .
rev . _ * d50 * ( 1994 ) 2225 ; see also e. iancu , a. leonidov , and l. mclerran hep - ph/0202270 for a review of the cgc and hep - ph/0202025 for a nice review of the cgc and rhic physics . | i review the present status of measurements at rhic and suggest possible upgrades which will be necessary to answer some of the most critical questions in relativistic heavy ion physics .
epsf 6.in 8.5 in 0.0 in 0.0 in 0.0 in * future measurements at rhic * richard seto university of california , riverside riverside , ca 92521 , usa _ proceedings of the nuclear dynamics workshop , nassau , bahamas , january 2002 _ |
for fifty years the metal - insulator transition has been one of the central themes@xcite of condensed matter physics .
the type we address here does not involve spatial disorder nor change of the number of charge carriers per cell ; the competing tendencies arise solely from the kinetic and potential energies in the hamiltonian , favoring itineracy and localization respectively , and the many real - material complexities that arise .
the classic categorization is that of the mott transition , treated in its most basic form with the single - band hubbard model . much has been learned about this model , but there are very few physical systems that are modeled faithfully by such a model .
real materials involve multiorbital atoms and thus extra internal degrees of freedom , and an environment that is often very active and may even react to the configuration of active sites .
mno is a transition metal monoxide ( tmo ) with open @xmath0 shell that qualifies as one of the simpler realizations of a prototypical , but real , mott insulator .
it is , certainly , a multiorbital system with the accompanying complexities , but the half - filled @xmath0 bands lead to a spherical , spin - only moment at ambient pressure . applying pressure to such
a system leads to a number of possibilities , including insulator - metal transition , moment reduction , volume collapse if a first - order transition ( electronic phase change ) occurs , and any of these may be accompanied by a structural phase transition , that is , a change in crystal symmetry .
the @xmath0 band width w of such a mott insulator is very susceptible to applied pressure , and is one of the main determining factors of the strength of correlation effects .
the `` closed subshell '' aspect makes mno an atypical @xmath0 monoxide , as shown for example by saito _
_ , who compiled@xcite effective parameters for this system from spectroscopic information .
an effective intra - atomic coulomb repulsion energy as defined by them , for example , is roughly twice as large as for the other @xmath0 monoxides .
the complexity of this compound can be considered in terms of the energy scales that are involved in the electronic structure and magnetism of these oxides .
these include the @xmath0 bandwidth w , an intra - atomic coulomb repulsion strength u , an intra - atomic @xmath3 exchange energy ( hund s rule j , or exchange splitting @xmath4 ) , the crystal field splitting @xmath5 , and the charge transfer energy @xmath6 ( the difference in mean mn @xmath0 and o @xmath1 site energies ) . in the magnetically ordered antiferromagnetic ( afm ) state , there are further symmetry lowering and ligand field subsplittings involving @xmath7 hybridization .
all of these scales change as the volume changes , making the pressure - driven mott transition a challenging phenomenon to describe .
although the objective of the current paper is to compare methods within the fcc ( rocksalt ) phase , it is useful first to recount what is known about the mott transition at this time . the current experimental information , mostly at room temperature , on the behavior of mno under pressure is summarized in fig .
[ mnophasediagram ] .
resistance measurements@xcite provided the first evidence of the mott transition in mno near @xmath8 gpa .
recent x - ray diffraction and emission spectroscopy measurements of the crystal structure and magnetic moment by yoo _
et al._@xcite have clarified the behavior .
around 90 gpa there is a structural transformation from the distorted b1 ( rocksalt ) phase to the b8 ( nias ) structure .
this structure change is followed at 105 gpa by the mott transition , consisting of a simultaneous volume collapse and moment collapse signifying a qualitative change in the electronic structure of the compound . on the theoretical side ,
little is known about how the mott transition occurs in a real multiband tmo in spite of the extensive studies of the mott transition in the single - band hubbard model , which has a simple spin - half moment at strong coupling and half - filling .
the numerous energy scales listed above , and the s=@xmath9 moment on mn arising from the five @xmath0 electrons , allow many possibilities for how the moment might disintegrate as the effective repulsion decreases .
the high pressure limit is clear : a nonmagnetic @xmath7 band metal in which kinetic energy overwhelms potential energy .
this is the competition studied in the ( simplified ) hubbard model .
the multiband nature has attracted little attention until recently , when for example the question of possible orbital - selective mott transitions@xcite have aroused interest .
one can imagine one scenario of a cascade of moment reductions s = @xmath10 before complete destruction of magnetism , as electrons use their freedom to flip spins ( as some competing energy overcomes hund s rule , for example ) .
in such a scenario there is the question of which orbital flips its spin at each spinflip , which involves a question of orbital selection and ordering .
at each flip the system loses exchange ( potential ) energy while gaining kinetic energy ( or correlation energy through ` singlet ' formation ) .
the manner in which kinetic energy changes is difficult to estimate because subband involvement means that there is no longer a single bandwidth w that is involved .
the increasing hybridization with o @xmath1 states under pressure strongly affects the kinetic energy , directly and through superexchange ( a kinetic energy effect ) .
it is known that conventional band theory ( local density approximation [ lda ] ) that does so well for so many materials gives poor results for @xmath0 monoxides in many respects , and some predictions are qualitatively incorrect ( _ viz .
_ no band gap when there should be a large gap of several ev ) .
thus even at the density functional level ( ground state energy , density , and magnetization ) some correlated approach is required . in the past fifteen years several approaches , which we refer to as correlated band theories , have been put forward , and each has had its successes in providing an improved description of some aspects of correlated tmos . although commonly called mean - field approaches with which they share many similarities , they are not mean - field treatments of any many - body hamiltonian .
rather , they are energy functionals based on the complete many - body hamiltonian , which must then be approximated due to limited knowledge of the exchange - correlation functional . in this paper , we provide a close comparison of certain results from four such methods : full potential lda+u , the hybrid exchange functional ( hse ) approach , the self - interaction - corrected local spin - density method ( sic - lsd ) , and a nonlocal pseudopotential - like variation of sic ( pseudo - sic ) . our main focus is to compare the predicted changes in energy , moment , and volume within the afmii rocksalt phase of mno . to keep the comparison manageable
we confine our attention to the rocksalt phase , since our emphasis is on comparison of methods and not yet the ultimate but more daunting task of modeling structural changes that may precede , or accompany , the mott transition .
rocksalt mno has an experimental equilibrium lattice constant @xmath11=4.45 ( cubic cell volume v@xmath12=88.1 @xmath13 ) .
density functional theory , like hartree - fock theory , deals in its most straightforward form with ground state properties , _
i.e. _ zero temperature .
the ground state is known to be the afmii phase in which @xmath14 layers have spins aligned , and successive layers are antiparallel .
the resulting symmetry is rhombohedral , with mn@xmath15 and mn@xmath16 being distinct sites ( although related through a translation + spin - flip operation ) .
thus , while most of the lore about transition metal monoxides is based upon cubic symmetry of the mn ( and o ) ion , in the ordered state the electronic symmetry is reduced .
it is obvious that individual wavefunctions will be impacted by this symmetry , _ viz . _
fourfold symmetry around the cubic axes is lost .
it has been emphasized by massidda _
et al._@xcite that zone - integrated , and even spin - integrated , quantities show the effects of this symmetry lowering ; for example , born effective charges lose their cubic symmetry .
since this issue arises in the interpretation of our results , we provide some background here . in cubic symmetry
the mn @xmath0 states split into the irreducible representations ( irreps ) denoted by @xmath17 and @xmath18 .
rhombohedral site symmetry results in the three irreducible representations @xmath19 , @xmath20 , and @xmath21 , the latter two being two fold degenerate .
the coordinate rotation from cubic to rhombohedral ( superscript @xmath22 and @xmath23 respectively ) is , with a specific choice for the orientation of the @xmath24 and @xmath25 axes in the rhombohedral system , @xmath26 applying this rotation of coordinates gives the @xmath0 orbitals in the rhombohedral frame in terms of those in the cubic frame ( @xmath27 ) : @xmath28 in rhombohedral coordinates it is useful to categorize the @xmath0 orbitals in terms of their orbital angular momentum projections along the rhombohedral axis : @xmath29
@xmath30 @xmath31 ; @xmath32 ; @xmath33 .
it is easy to see that @xmath34 specifies groups of states that only transform into combinations of themselves under trigonal point group operations .
note that the unique @xmath19 symmetry state in rhombohedral coordinates is the fully symmetric combination of the cubic @xmath17 states .
the other two irreps are both @xmath18 doublets .
while @xmath35 and @xmath36 form representations of these irreps , if there are components of the crystal field that are not diagonal in the l=(2,@xmath37 ) basis , these states will mix . then each of the resulting ( orthonormal )
irreps @xmath20 , and @xmath21 will contain both @xmath35 and @xmath36 components .
such mixing does occur in mno and complicates the symmetry characterization of the @xmath0 states .
for lda band structure plot ( fig .
[ ldabands ] ) we used version 5.20 of the full - potential local orbital band structure method ( fplo@xcite ) .
relativistic effects were incorporated on a scalar - relativistic level .
we used a single numerical basis set for the core states ( mn @xmath38 and o @xmath39 ) and a double numerical basis set for the valence sector including two @xmath40 and @xmath0 radial functions , and one @xmath41 radial function , for mn , and two @xmath42 and @xmath1 radial functions , and one @xmath0 radial function , for o. the semi core states ( mn @xmath43 ) are treated as valence states with a single numerical radial function per @xmath44-shell .
the local density exchange - correlation functional pw92 of perdew and wang@xcite was used .
the lda+u approach of including correlation effects is to ( 1 ) identify the correlated orbital , @xmath0 in this case , ( 2 ) augment the lda energy functional with a hubbard - like term ( coulomb repulsion u ) and hund s ( exchange j ) energy between like spins , ( 3 ) subtract off a spin - dependent average of this interaction energy to keep from double - counting repulsions ( once in lda fashion , once in this u term ) , and ( 4 ) include the correlated orbital occupation numbers in the self - consistency procedure , which leads to an orbital - dependent hartree - fock - like potential acting on the correlated orbitals .
the addition to the energy functional has the schematic form @xmath45 . \label{ldau}\end{aligned}\ ] ] we actually use the coordinate - system independent form of lsda+u @xcite implemented in fplo,@xcite which leads to four @xmath46 indices on @xmath47 and @xmath48 which for simplicity have not been displayed ( nor has the full off - diagonal form of the occupation matrices @xmath49 ) .
this treatment of the on - site interactions @xmath47 and @xmath48 incorporates on - site correlation effects in the mn 3d - shell .
we have used the so called ` atomic - limit ' ( strong local moment ) form of the double - counting correction , the last term in eq .
this form is appropriate for the high - spin state , but it is less obviously so for the low - spin state that is found at reduced volumes .
the slater parameters were chosen according to @xmath50 , @xmath51 and @xmath52 . + the shape of the basis orbitals has been optimized yielding a sufficient accuracy of the total energy over the range of geometries considered in this work .
the @xmath53-integrals are performed via the tetrahedron method with an irreducible mesh corresponding to 1728 ( @xmath54 ) points in the full brillouin zone .
the sic - lsd method addresses the unphysical self - interaction in the lda treatment of localized states .
itinerant states , being spread over space without finite density in any given region , do not experience this self - interaction within the lda treatment .
should there be localized states , confined to some region and giving a finite density , they will suffer an unphysical self - interaction in the lsd method .
this issue then clearly arises in the itinerant - localized transition in mno and other correlated systems .
the basic premise of the sic - lsd method is that localized electrons should experience a different potential from that of itinerant electrons,@xcite analogous to that of an atomic state whose self - interaction must be removed .
then electrons on the surrounding atoms are allowed to accommodate self - consistently .
this distinction of localized versus itinerant state is addressed in sic - lsd by extending the energy functional in the form @xmath55 + e_{xc}^{lsd}[n_{\alpha}].\end{aligned}\ ] ] here @xmath56 $ ] represent the hartree ( classical coulomb ) energy of a density @xmath57 .
the self - coulomb energy @xmath58 $ ] and self - exchange - correlation @xmath59 $ ] energies are subtracted off for each localized state @xmath60 with density @xmath61 . whether states are localized or not ( with non - zero , respectively zero self - interaction ) is determined by minimization of this functional , allowing localized as well as itinerant states @xmath60 .
since the correction vanishes for itinerant states , the sum finally includes only the self - consistently localized states .
the localized and itinerant states are expanded in the same basis set , and minimization becomes a process of optimizing the coefficients in the expansion of the states ( as other band structure methods do , except that bloch character is imposed in other methods ) .
the implementation of temmerman and collaborators@xcite used here incorporates the atomic - sphere approximation ( asa ) of the linear muffin - tin orbital electronic structure method@xcite ( lmto ) in the tight - binding representation.@xcite further details can be found in ref . . the large computing requirements ( compared to lda ) of the sic - lsd method , even for materials with small unit cells ,
has led to an alternative approach @xcite , in which the self - interaction part of the kohn - sham potential is approximated by a non - local , atomic - like contribution included within the pseudopotential construction .
the original implementation of this scheme has given important improvements over lsda results for non - magnetic ii - vi and iii - v semiconductors , but was not applicable to metals or to magnetic and highly - correlated systems where there is a coexistence of strongly localized and hybridized electron charges .
the pseudopotential self - interaction corrected calculations presented here were performed using the recently developed `` pseudo - sic '' method of filippetti and spaldin @xcite .
this pseudo - sic approach represents a compromise between the fully self - consistent implementations of svane _ et al .
_ @xcite and the alternative method of vogel et al .
@xcite , in that the sic calculated for the atom ( as in ref .
@xcite ) is scaled by the electron occupation numbers calculated self - consistently within the crystal environment .
this allows the sic coming from localized , hybridized , or completely itinerant electrons to be discriminated , and permits the treatment of metallic as well as insulating compounds , with minimal computational overhead beyond the lsda . in this pseudo - sic procedure ,
the orbital sic potential is taken from the isolated neutral atom and included in the crystal potential in terms of a nonlocal projector , similar in form to the nonlocal part of the pseudopotential .
the bloch wave functions are projected onto the basis of the pseudo - atomic orbitals , then , for each projection , the potential acting on the bloch state is corrected by an amount corresponding to the atomic sic potential .
note that , within this formalism , a physically meaningful energy functional which is related to the kohn - sham equations by a variational principle is not available .
however , a suitable expression for the total energy functional was formulated in ref .
@xcite and shown to yield structures in good agreement with experiment .
we use this functional here to calculate the bondlength dependence of the total energy .
we have used ultra - soft pseudopotentials with an energy cutoff of 35 ry .
an @xmath62 monkhorst - pack grid was used for @xmath63-point sampling .
the low - spin and high - spin solutions were obtained by setting initial magnetization to 5@xmath64 or 1@xmath64 respectively .
+ the hybrid - exchange dft approximation mixes a fraction of the exact , non - local , exchange interaction ( which uses the hartree - fock [ hf ] expression ) with the local , or semi - local , exchange energy of the lda or the generalized gradient approximation ( gga ) .
the pbe0 functional takes the form : @xmath65 where @xmath66 and @xmath67 are the exchange and the pbe gga functionals . the mixing parameter @xmath68 = 1/4
was determined via perturbation theory [ 21 ] and @xmath69 is the pbe correlation energy . in this work
we use the hybrid method recently developed by heyd , scuseria and enzerhof ( hse)).@xcite it is based upon the pbe0 functional , but employs a screened , short - range ( sr ) hartree - fock ( hf ) exchange instead of the full exact exchange , which results in a more efficient evaluation for small band gap systems . in this approach , the coulomb operator is split into short - range ( sr ) and long - range ( lr ) components respectively @xmath70 where @xmath71 is a parameter that can be adjusted for numerical or formal convenience .
the expression for the hse exchange - correlation energy is @xmath72 where e@xmath73(@xmath71 ) is the sr hf exchange computed for the sr part of the coulomb potential , e@xmath74(@xmath71 ) and e@xmath75(@xmath71 ) are the sr and the lr components of the pbe exchange , respectively .
the cited papers should be consulted for further details .
the hse functional has been found to yield results in good agreement with experiment for a wide range of solids and molecules @xcite .
this functional is implemented in the development version of the gaussian quantum chemistry package @xcite .
we use the towler basis@xcite of gaussian functions for our basis set .
it consists of a [ 20s12p5d/5s4p2d ] contraction for mn and a [ 14s6p/4s3p ] basis for o , optimized for hf studies on mno .
a pruned grid for numerical integration with 99 radial shells and 590 angular points per shell was used .
the @xmath76-space was sampled with a @xmath77 mesh .
low spin and high spin antiferromagnetic initial guesses were obtained using the crystal field approach by patching the density matrix obtained from diagonalization of the harris functional @xcite with the density matrices obtained from calculation on ions in the appropriate ligand field @xcite .
the origin , and the proper description , of the moments and the band gaps in transition metal monoxides have a long history . the earliest question centered on the connection between the antiferromagnetic ( afm ) order and the insulating behavior .
band picture@xcite could account in a one - electron manner for a gap arising from afm order , whereas mott s picture of correlation - induced insulating behavior@xcite was a many - body viewpoint with insulating behavior not connected to the magnetic order .
the proper general picture in these monoxides arose from studies of transport above the nel temperature and with introduction of defects , giving them the designation as mott insulators .
much progress on the understanding of mno and the other monoxides came from early studies using lda . while understanding that lda does not address the strong correlation aspect of the electronic structure , mattheiss@xcite and terakura _
et al._@xcite quantified the degree and effects of @xmath7 interactions , and pointed out the strong effect of magnetic ordering on the band structure .
more recently , pask and collaborators@xcite have studied the structural properties , and the rhombohedral distortion , with lda and gga approximations .
the symmetry lowering and resulting structure is described well , and in addition they found that afm ordering results in significant charge anisotropy .
effects of afm order were further probed by posternak _
et al . _ by calculating and analyzing maximally localized wannier functions for the occupied states.@xcite the application of correlation corrections in mno already has a colorful history . the first work , by svane and gunnarsson@xcite and by szotek _
et al._,@xcite was in the application of the sic - lsd method .
the former pair correctly obtained that mno , feo , coo , nio , and cuo are afm insulators , while vo is a metal .
they calculated a gap of 4 ev for mno .
_ used a fairly different implementation of the sic - lsd approach but find a similar gap ( 3.6 ev ) .
their @xmath0 states lay about 6 ev below the center of the @xmath1 bands , although hybridization was still clearly present . in this same time frame ,
anisimov , zaanen , and andersen introduced@xcite the lda+u method with application to the transition metal monoxides .
they obtained a band gap of 3.5 ev but few other results on mno were reported .
kotani implemented@xcite the `` dft exact - exchange '' method of talman and shadwick@xcite to crystal calculations .
this method consists of taking the fock expression for the exchange energy in the dft functional , then performing a kohn - sham solution ( minimization ) , giving a local exchange potential ( `` optimized effective potential '' ) . in kotani
s results for mno , the mn @xmath18 and @xmath17 bands form very narrow ( almost atomic - like ) bands between the occupied o @xmath1 bands and the conduction bands . takahashi and igarashi@xcite proposed starting from the hartree - fock exchange and adding correlation from a local , three - body scattering viewpoint .
their corrections were built on a parametrized tight - binding representation , and they obtained small self - energy corrections for mno , much smaller than they obtained for the other transition metal monoxides . the effective potential approach used by kotani
was extended by solovyev and terakura@xcite in an unconventional way .
they obtained an effective potential using the criterion that it had to reproduce the spin - wave spectrum , _ i.e. _ that it had to describe the magnetic interactions correctly .
they found clear differences when comparing to the lda+u and the optimized effective potential results , and discussed limitations of the one - electron band method itself .
more recently , savrasov and kotliar applied a dynamical extension@xcite of the lda+u method ( dynamical mean field theory ) to mno and nio .
being a self - energy method , this is not really a correlated band theory .
for the properties they calculated ( band gap , effective charges , dielectric constant , optic phonon frequencies ) the dynamical results are similar to the lda+u results and differ considerably from lda values .
even though hybrid - exchange dft applications to solids are still in their infancy , there have been two previous studies of mno .
the first , by bredow and gerson,@xcite utilized the b3lyp hybrid functional . unlike the lda and gga , they found b3lyp provided an excellent band gap for mno .
more recently , franchini _
et al . _ have examined mno in more detail using the pbe0 approximation.@xcite they also found a gap , lattice constant and density of states in quite good agreement with experiment . in particular , the distorted db1 rhombohedral structure was determined to be the minimum energy geometry , in agreement with experiment .
neither the b3lyp nor the pbe0 approximation can be applied to the metallic side of the transition of interest here . for that
, we must turn to the screened hybrid - exchange of hse . therefore , while there has been thorough lda studies of mno and a variety of approaches to treatment of the correlation problem , nearly all of these have considered only ambient pressure or small variations of volume near zero pressure .
the work described in the following sections focuses on testing the four different correlated band methods from ambient conditions to high pressures , through the volume collapse regime , to see whether some basic foundation can be laid for the understanding and theoretical description of pressure - driven mott transitions in real materials .
our principal results revolve around the first - order transition , high@xmath78low volume which is also high@xmath78low moment in nature . for convenience
we use from here on the specific volume @xmath79 , the volume referenced to the experimental zero - pressure volume .
the equations of state have been fitted for both high volume and low volume phases for each computational method , and the resulting constants are presented and analyzed below .
.[lda_table ] local density approximation results obtained from the four codes used in this work , for the equilibrium lattice constant relative to the experimental value ( @xmath80 ) , the bulk modulus b , the energy gap , and mn moment .
the experimental values for b fall in the range 142 - 160 gpa ( see caption to table [ data_table ] ) .
all calculations are carried out for the antiferromagnetic ii phase of mno in the cubic ( nacl ) structure .
differences are due to ( 1 ) differing exchange - correlation functions that are used , ( 2 ) approximations made in algorithms to solve the kohn - sham equations and evaluate the energy , ( 3 ) basis set quality , and ( 4 ) for the moment , the operational definition of the `` mn moment '' differs .
the notation follows .
fplo : full potential local orbital method with perdew - wang ( 1992 ) exchange - correlation functional .
gaussian : gaussian local orbital code , slater exchange ( @xmath81=2/3 ) and vosko - wilk - nusair correlation .
pw - ussp : planewave basis using ultrasoft pseudopotentials , with the perdew - zunger exchange - correlation functional .
lmto - asa : linear muffin - tin orbital code in the atomic sphere approximation , using the perdew - zunger exchange - correlation functional . [ cols="^,^,^,^,^ " , ] the equation of state ( eos ) energy vs. volume curves for the various functionals are collected in fig .
[ evsv ] . for each correlated band method
a large volume , high - spin state and a small volume , low - spin state are obtained .
the analysis to obtain the first - order volume collapse transition was done as follows . for each high volume and low volume phase
separately , an eos function @xmath82 was determined ( @xmath83=high , low ) by a fit to the murnaghan equation .
both fits give minima , with the most relevant one being for the high spin phase and being the predicted equilibrium volume @xmath84 .
the pressure is obtained from the volume derivative of the eos , which is inverted to get @xmath85 . equating the enthalpies
@xmath86 + pv(p)$ ] of the two phases gives the critical pressure @xmath87 .
the volumes at this pressure then give the volume collapse @xmath88 = @xmath89 .
the various quantities for all four computational schemes are given in table [ data_table ] , along with the uncorrelated results of cohen _
et al._@xcite not surprisingly given the other differences that will be discussed , there are substantial variations among the critical pressures and related quantities .
particularly noticeable already in the eos is the result that the energy difference between the low - spin energy minimum and the high - spin one is @xmath900.2 ev for the lda+u and pseudo - sic methods , while the hse method gives roughly twice the energy difference ( 0.4 ev ) , and the sic - lsd method gives roughly 0.6 ev .
we now mention other noteworthy features of the calculated data in table [ data_table ] .
+ ( 1 ) the predicted equilibrium volume from the lda+u method is the smallest of the four methods ( @xmath91=0.93 ) , thus overbinding .
the hse value is almost indistinguishable from the observed value , while the sic - lsd and pseudo - sic methods give underbinding ( @xmath91 = 1.04 , 1.09 respectively ) .
+ ( 2 ) the pseudo - sic method predicts the transition to occur at a relatively small volume reduction ( @xmath92=0.86 ) ; the other methods give the onset of transition at @[email protected] .
+ ( 3 ) the critical pressure p@xmath95=56 gpa predicted by pseudo - sic is smallest of the methods .
p@xmath96 in lda+u ( 123gpa ) is comparable to that of lda ; those of sic - lsd and hse are higher ( 204 , 241 gpa , respectively ) .
+ ( 4 ) the sic - lsd method predicts a transition to a low volume phase that is much softer than the high volume phase , a phenomenon that is extremely unusual in practice but not disallowed .
there are two possible sources of this difference : ( i ) in sic - lsd the system becomes completely lda - like in the low volume phase whereas in the other methods the @xmath0 states are still correlated , or ( ii ) the lmto - asa method involves approximations that pose limitations in accuracy.@xcite + ( 5 ) the values of @xmath97 in the large volume phase vary although not anomalously so , given the differences discussed just above .
+ ( 6 ) the values of @xmath98 in the high volume phase are reasonably similar across the methods : @xmath99 .
the variation in the collapsed phase is greater .
the moment collapse behavior of each method is collected in fig .
[ mvsv ] . for comparison , the gga result presented by cohen _
et al._@xcite was a moment collapse from 3.4@xmath64 to 1.3@xmath64 at the volume given in table [ data_table ] .
on the broadest level , the predictions for the mn moment show
_ remarkable similarity_. at low pressure , all methods of course give the high - spin s=@xmath9 configuration of the mn@xmath100 ion , with the local moment being reduced slightly from 5@xmath64 by @xmath7 mixing .
this electronic phase persists over a substantial volume reduction , giving way in all cases to an s=@xmath101 state , _ not _ the nonmagnetic @xmath102 result that might naively be anticipated .
three methods give a stable moment very near 1@xmath64 over a range of volumes .
the sic - lsd method ( which in the collapsed phase is simply lda ) is alone in giving a varying moment , one that reduces from 1.4 @xmath64 at @xmath103 down to 0.5 @xmath64 at @xmath104 one difference between the methods lies in how soon the low - spin state becomes metastable , _
i.e. _ when it is possible to obtain that state self - consistently , as opposed to when it becomes the stable solution ( which was discussed in the eos subsection )
. the state is obtained already at ambient volume in the hse method ; the pseudo - sic method obtains the low - spin state just below @xmath105 ; for the lda+u method , it was not followed above @xmath103 .
it should be emphasized however that no concerted measures were taken to try to follow all solutions to the limit of their stability . in fig .
[ gapvsa ] the calculated band gap of both high - spin and low - spin states are shown for all methods . here
the behavior differs considerably between the methods , in part because at certain volumes the gap lies between different bands for some of the methods . at ambient pressure
the pseudo - sic and hse methods obtain a gap of 3.5 - 4 ev , while that for sic - lsd is 2.9 ev , and that of the lda+u method is even lower , less than 2 ev .
experimental values lie in the 3.8 - 4.2 ev range . referring to fig .
2 , it can be observed that the large volume gap depends on the position of the majority @xmath0 states with respect to the free - electron - like band , _
i.e. _ it is not the @xmath106 mott gap . both of the former approaches show only a slight increase as pressure is applied , reaching a maximum around @xmath107 where a band crossing results in a decreasing gap from that point . for pseudo - sic
, there is an almost immediate collapse to a metallic low spin state . within hse
, mno collapses to the low - spin state at a volume @xmath108 , coincident with the closing of the gap in the low - spin state .
the sic - lsd and lda+u gaps , smaller initially , show a much stronger increase with pressure , and also incur the band crossover that leads to decrease of the gap ( within the high - spin state ) .
in this section we analyze the character of the states just above and just below the mott transition , as predicted by each of the methods . due to the differing capabilities of the codes
, the quantities used for analysis will not be identical in all cases . in the figs .
6 - 9 we present for uniformity the dos in the high volume phase at the equilibrium volume ( @xmath109 ) and in the collapsed phase at @xmath110 ( @xmath93=0.6 ) .
note ( from table i ) that this specific volume does not correspond to any specific feature in the phase diagram for any method , although it lies in the general neighborhood of the volume at the collapse .
changes within the collapsed phase are continuous , however , so the plots at @xmath110 are representative of the collapsed phase .
the projected doss ( pdoss ) in fig . [ ldaudos ] refer to projections onto mn @xmath0 orbitals , with the @xmath111-axis being the rhombohedral axis , the @xmath19 @xmath112 ( @xmath113=0 ) state ; the @xmath114 pair \{@xmath115 } ( @xmath113=1 ) ; and the @xmath18 pair \{@xmath116 } ( @xmath113=2 ) .
because the two @xmath18 representations have the same symmetry , they can mix and the actual combinations @xmath117 are orthogonal linear combinations of @xmath118 which depend on interactions .
for the lda+u results , however , there is little mixing of the @xmath118 pairs .
the character of the transition is simple to describe : the @xmath114 pair ( @xmath113=1 with respect to the rhombohedral axis ) simply flips its spin .
this s=@xmath101 state is unexpected and quite unusual .
first , each @xmath0 orbital is still singly occupied , verified by plotting the charge density on the mn ion and finding it just as spherical as for the high - spin state .
second , each @xmath0 orbital is essentially fully spin - polarized , with the configuration being @xmath119 .
a plot of the spin density@xcite reveals the unanticipated strong anisotropy with nodal character , characteristic of spin - up @xmath120 and @xmath113=2 orbitals , and spin - down @xmath113=1 orbitals ( in the rhombohedral frame ) .
third , it makes this transition with essentially zero change in the gap , which is 3.5 ev .
the band structure changes completely , however , so the close similarity of the gaps on either side is accidental . in the high volume phase the distribution and overall width of the occupied @xmath0 states , shown in fig .
[ hsedos ] , is similar to that of the lda+u method ( previous subsection ) .
the gap is larger , as discussed earlier .
the collapsed phase shows new characteristics .
the gap collapses from 4 ev at @xmath109 to essentially zero at the onset of the collapsed phase at @xmath93=0.55 , making this an insulator - to - semimetal transition .
the metallic phase then evolves continuously as the pressure is increased beyond @xmath87 .
the @xmath0 configuration can be characterized as @xmath121 , resulting in a moment of 1@xmath64 .
the corresponding spin density is strongly anisotropic , although in a different manner than is the case for lda+u . the spin - decomposed spectrum from this method is shown in fig .
[ psicdos ] , symmetry - projected as done above for the lda+u method .
the pdoss in the high - spin state are quite similar to those given by the lda+u method
. the transition could hardly be more different , however .
the gap collapses in an insulator - to - good - metal character , the fermi level lying within both majority and minority bands .
the majority bands are the @xmath114 pair ( @xmath113=1 ) and are only slightly occupied . in the minority bands
both @xmath18 and @xmath114 are roughly quarter filled .
the reason this solution is ( at least locally ) stable seems clear : e@xmath122 falls in a deep valley in the minority dos . in the collapsed , low - spin state , the @xmath19 orbital of both spins
is occupied , and the majority @xmath18 pair is also fully occupied .
this results in a configuration that can be characterized roughly as @xmath123 , giving spin @xmath124 thus the fact that the same moment is found in the low - spin state as was found with the lda+u and pseudo - sic methods seems accidental , because in those methods the energy gap required integer moment whereas the pseudo - sic solution is firmly metallic .
it is in fact close to half metallic , which accounts for the near - integer moment . in pseudo - sic ,
the @xmath19 orbital is unpolarized ( spin - paired ) , the @xmath18 pair \{@xmath115 } ( @xmath113=1 ) is positively polarized , and the @xmath114 pair \{@xmath116 } ( @xmath113=2 ) is negatively polarized but to a smaller degree .
this method give much more tightly bound @xmath0 state in the high - spin state than the other methods .
an associated feature is that the majority - minority splitting , the `` effective u''@xmath125 plus the exchange splitting 13 ev , more than twice as la large as used in the lda+u method ( both in this paper and elsewhere ) .
note that in the sic - lsd method @xmath126 is a true slater self - coulomb integral , whereas in the lda+u method the value of @xmath47 represents the ( somewhat screened ) coulomb interaction between a @xmath0 electron and an _ additional _
@xmath0 electron , so agreement between the two is not expected .
nevertheless the difference is striking . all five majority @xmath0 states
are localized , leading to the self - interaction potential that binds them .
the majority @xmath0 states lie 6 ev below the center of the @xmath1 bands and hybridize very weakly , which accounts for the very narrow , almost core - like @xmath0 bands . in the collapsed phase
, there are no localized states and the usual lsda results re - emerges .
all @xmath0 states make some contribution to the moment , but the strongest contribution arises from the @xmath19 orbital .
one might argue in the lda+u result that the net moment arises from the @xmath19 orbital ( with the moments arising from the @xmath18 and @xmath114 orbital canceling ) .
however , the electronic structure in the collapsed phase of this sic - lsd method is very different from that of the lda+u method .
the results of four correlated band methods
lda+u , sic - lsd , pseudo - sic , and hse have been compared for the equation of state ( for both normal and collapsed phases ) , electronic structure ( including @xmath0 configuration ) , and the mn moment under pressure . in order to make the comparison as straightforward as possible ,
the crystal structure was kept cubic ( rocksalt ) . to compare seriously with experiment
, one must account for the coupling of the afm order to the structure , because this results in a substantial rhombohedral distortion of the lattice.@xcite then structural transitions ( particularly to the b8 phase ) must also be considered . the large volume , high - spin phases are qualitatively the same for the various functionals : afm with a fully - polarized @xmath127 configuration . due to the large charge transfer energy ,
the configuration remains @xmath128 at all volumes studied here .
the predicted equations of state ( which give the equilibrium volume , bulk modulus , and its pressure derivative ) show rather strong variation , suggesting that the extension of the mn @xmath0 and o @xmath1 functions , or their hybridization , differ substantially even in this large volume phase .
of course , since the functionals are different , any given density would lead to different energies . under pressure ,
the gap initially increases ( all methods give this behavior ) , and the system suffers a first - order transition ( isostructural , by constraint ) to a collapsed phase where hybridization must be correspondingly stronger . uniformly among the methods , the moment collapse reflects an s=@xmath9 to s=@xmath101 transition , rigorously so for the lda+u method for which the collapsed phase retains a gap , and less rigorously so for the the other methods where the collapsed phase is metallic .
it is remarkable that none of the methods gives a collapse to a nonmagnetic state , which probably would be the most common expectation .
this s=@xmath129 moment collapse is related in some cases to the local symmetry of the mn @xmath0 orbitals in the afm phase ( being most obvious for the lda+u results of fig . [ ldaudos ] ) .
the symmetry is @xmath130 , _
i.e. _ a singlet and two doublets per spin direction . without further symmetry - breaking ( orbital ordering )
an s=@xmath131 transition requires a single spin flip , which could only be the @xmath19 spin .
however , the @xmath19 state is more tightly bound than as least one of the two doublets both for lda+u ( fig .
[ ldaudos ] ) and for pseudo - sic ( fig .
[ psicdos ] ) . in lda+u
the @xmath18 doublet flips its spin , while in pseudo - sic the @xmath19 singlet flips its spin leaving the minority @xmath18 and @xmath114 doublets partially occupied and therefore metallic .
this symmetry - related behavior depends of course on the magnetic ordering that gives rise to the ( electronic ) rhombohedral symmetry . above the nel temperature ,
the moment collapse at the mott transition may proceed differently because the mn moment would lie at a site of ` cubic ' symmetry ( a dynamic treatment could include the effect of short range spin correlations ) .
our study provides some of the first detailed information on how magnetic moments in a real material may begin to disintegrate without vanishing identically , at or near a mott transition , when correlation is taken into account .
it is accepted that dynamic processes will be required for a truly realistic picture of the mott ( insulator - to - metal ) transition . however , a moment collapse between two insulating phases ( as described here by two of the methods ) may be described reasonably by a correlated band ( static ) approach . beyond this similar amount of moment collapse ,
the four functionals give substantially different collapsed phases : differences in the mn @xmath0 magnetic configuration ( although all remain @xmath128 ) and differences in conducting versus insulating behavior .
it is not surprising therefore that the collapsed - phase equations of state differ considerably between the methods .
the differences in predictions can be traced , in principle , to the different ways in which exchange and correlation are corrected with respect to lda .
one clear shortcoming of lda is in the local approximation to the exchange energy .
the hse method deals with this problem directly , by using 25% hartree - fock exchange .
the self - interaction of the sic - lsd method is largely a self - exchange energy correction , subtracting out the spurious self - coulomb energy that occurs in the hartree functional if an orbital chooses to localize .
a self-(local density ) correlation correction is also included in sic - lsd .
the pseudo - sic method includes the same correction if applied to an atom , but in a crystal the pseudo - sic energy correction and change in potential takes a substantially different form , as the difference in predictions reflects .
the lda+u method is rather different in this respect : it specifically does not subtract out any self - interaction ( although it is sometimes discussed in this way ) . in the form
[ ldau ] of lda+u correction , the second term is simply an lda - like average of the first term .
the on - site coulomb repulsion is treated hartree - fock like , leading to an orbital - dependent , occupation - dependent potential .
each method has its own strengths , and each is only an anticipated improvement on lda toward a better , more general functional .
it is expected that more details of the results may be published separately by the respective practitioners .
d.k . , j.k . and w.e.p .
acknowledge support from department of energy grant de - fg03 - 01er45876 .
r.l.m . and c.v.d .
thank the doe bes heavy element chemistry program and lanl ldrd for support .
the work at rice university was supported by doe grant de - fg02 - 01er15232 and the welch foundation .
we acknowledges important interactions within , and some financial support from , the department of energy s stewardship science academic alliances program .
t.c.s . were supported by the office of basic sciences , u.s .
department of energy .
n.s . was supported by the national science foundation s division of materials research information technology research program , grant no .
dmr-0312407 , and made use of mrl central facilities supported by the national science foundation award no .
dmr-05 - 20415 .
this collaboration was stimulated by , and supported by , doe s computational materials science network .
m. imada , a. fujimori , and y. tokura , rev .
phys . * 70 * , 1039 ( 1998 ) .
t. saitoh , a. e. bouquet , t. mizokawa , and a. fujimori , phys .
b * 52 * , 7934 ( 1995 )
. j. r. patterson , c. m. aracne , d. d. jackson , v. malba , s. t. weir , p. a. baker , and y. k. vohra , phys .
b * 69 * , 220101 ( 2004 ) . c. s. yoo , b. maddox , j .- h .
p. klepeis , v. iota , w. evans , a. mcmahan , m. hu , p. chow , m. somayazulu , d. husermann , r. t. scalettar and w. e. pickett , phys .
. lett . * 94 * , 115502 ( 2005 ) .
y. syono , riken rev .
* 27 * , 72 ( 2000 ) .
r. d. cowan , phys . rev . * 163 * , 54 ( 1967 ) .
i. lindgren , intl .
. chem . * 5 * , 411 ( 1971 ) .
j. p. perdew and a. zunger , phys .
b * 23 * , 5048 ( 1981 ) .
w. m. temmerman , a. svane , z. szotek , and h. winter , in _ electronic density functional theory : recent progress and new directions _ , edited by j. f. dobson , g. vignale , and m. p. das ( plenum , new york , 1998 ) . o. k. andersen , phys . rev .
b * 12 * , 3060 ( 1975 ) .
o. k. andersen and o. jepsen , phys .
* 53 * , 2571 ( 1984 ) .
a. svane , w. m. temmerman , z. szotek , j. laegsgaard , and h. winter , intl .
. chem . * 77 * , 799 ( 2000 ) .
d. vogel , p. krger , and j. pollman , phys .
b * 54 * , 5495 ( 1996 ) .
filippetti , a. and spaldin , n. a. , phys .
b , * 67 * , 125109 ( 2003 ) . a. svane and o. gunnarsson , phys .
lett . * 65 * , 1148 ( 1990 ) .
j. heyd , g. scuseria , and m. ernzerhof , j. chem .
* 118 * , 8207 ( 2003 ) ; _ ibid . _
* 124 * , 219906 ( 2006 ) .
j. heyd and g. scuseria , j. chem .
* 120 * , 7274 ( 2004 ) .
j. heyd and g. scuseria , j. chem .
phys . , * 121 * , 1187 ( 2004 ) .
j. p. perdew , m. ernzerhof , and k. burke , j. chem . phys . * 105 * , 9982 ( 1996 ) . m. j. frisch @xmath132 , gaussian development version , revision c.01 .
gaussian , inc . ,
wallingford , ct ( 2004 ) .
m. d. towler , n. l. allan , n. m. harrison , v. r. saunders , w.c .
mackrodt , and e. apra , phys .
b * 50(8 ) * , 5041 ( 1994 ) .
k. d. heath , w. c. mackrodt , v. r. saunders , and mauro causa , j. mater . chem .
* 4(6 ) * , 825 ( 1994 ) .
m. d. towler , n. l , allan , n.m .
harrison , v. r. saunders , and w. c. mackrodt , j. phys . : condens .
matter * 7 * , 6231 , ( 1995 ) . j. c. slater , phys .
rev . * 82 * , 538 ( 1951 ) .
n. f. mott , proc .
a * 62 * , 416 ( 1949 ) ; _ metal - insulator transitions _
( taylor and francis , london , 1990 ) . l. f. mattheiss , phys .
b * 5 * , 290 , 306 ( 1972 ) .
k. terakura , t. oguchi , a. r. williams , and j. kbler , phys .
b * 30 * , 4734 ( 1984 ) .
j. e. pask , d. j. singh , i. i. mazin , c. s. hellberg , and j. kortus , phys .
b * 64 * , 024403 ( 2001 ) .
m. posternak , a. baldereschi , s. massidda , and n. marzari , phys .
b * 65 * , 184422 ( 2002 )
. z. szotek , w. m. temmerman , and h. winter , phys .
b * 47 * , 4029 ( 1993 ) .
v. i anisimov , j. zaanen , and o. k. andersen , phys . rev .
b * 44 * , 943 ( 1991 ) .
t. kotani , phys .
rev . lett . * 74 * , 2989 ( 1995 ) .
t. kotani and h. akai , physica b * 237 * , 332 ( 1997 ) .
t. kotani , j. phys . : condens .
matt . * 10 * , 9241 ( 1998 ) .
j. d. talman and w. f. shadwick , phys .
a * 14 * , 36 ( 1976 ) .
m. takahashi and j. igarashi , phys .
b * 54 * , 13566 ( 1996 ) . i. v. solovyev and k. terakura , phys . rev .
b * 58 * , 15496 ( 1998 ) .
j. zhang , phys .
minerals * 26 * , 644 ( 1999 ) .
r. e. cohen , i. i. mazin , and d. g. isaak , science * 275 * , 654 ( 1997 ) .
z. fang , i. v. solovyev , h. sawada , and k. terakura , phys .
b * 59 * , 762 ( 1999 ) . | the electronic structure , magnetic moment , and volume collapse of mno under pressure are obtained from four different correlated band theory methods ; local density approximation + hubbard u ( lda+u ) , pseudopotential self - interaction correction ( pseudo - sic ) , the hybrid functional ( combined local exchange plus hartree - fock exchange ) , and the local spin density sic ( sic - lsd ) method .
each method treats correlation among the five mn @xmath0 orbitals ( per spin ) , including their hybridization with three o @xmath1 orbitals in the valence bands and their changes with pressure .
the focus is on comparison of the methods for rocksalt mno ( neglecting the observed transition to the nias structure in the 90 - 100 gpa range ) .
each method predicts a first - order volume collapse , but with variation in the predicted volume and critical pressure .
accompanying the volume collapse is a moment collapse , which for all methods is from high - spin to low - spin ( @xmath2 ) , not to nonmagnetic as the simplest scenario would have .
the specific manner in which the transition occurs varies considerably among the methods : pseudo - sic and sic - lsd give insulator - to - metal , while lda+u gives insulator - to - insulator and the hybrid method gives an insulator - to - semimetal transition .
projected densities of states above and below the transition are presented for each of the methods and used to analyze the character of each transition . in some cases the rhombohedral symmetry of the antiferromagnetically ordered phase clearly influences the character of the transition . |
the aim of this paper is to study the dispersive properties of the linear wave equation on real hyperbolic spaces and their application to nonlinear cauchy problems .
this theory is well established for the wave equation on @xmath0 : @xmath1 for which the following strichartz estimates hold : @xmath2 on any ( possibly unbounded ) interval [email protected] mm , under the assumptions that @xmath4 and separately the couples .25mm@xmath5{\hskip-.25mm}\times{\hskip-.25mm}[{\hskip.25mm}2,2\,\frac{n-1}{n-3})$ ] .25mmsatisfy the admissibility conditions @xmath6 the estimate ( [ eqr3 ] ) holds also at the endpoint @xmath7 when @xmath8 .
when @xmath9 this endpoint is @xmath10 and the estimate ( [ eqr3 ] ) fails in this case without additional assumptions ( see @xcite and @xcite for more details ) .
these estimates yield existence results for the nonlinear wave equation in the euclidean setting .
the problem of finding minimal regularity on initial data ensuring local well
posedness for semilinear wave equation was addressed for higher dimensions and nonlinearities in @xcite , and then almost completely answered in @xcite .
once the euclidean case was more or less settled , several attemps have been made in order to establish strichartz estimates for dispersive equations in other settings .
here we consider real hyperbolic spaces .25mm@xmath11 , which are the most simple examples of noncompact riemannian manifolds with negative curvature . for geometric reasons
, we expect better dispersive properties hence stronger results than in the euclidean setting .
it is well known that the spectrum of the laplace
beltrami operator .25mm@xmath12 on @xmath13 is the half
line [email protected] mm , where .25mm@xmath15 .
thus one may study either the _ non shifted _ wave equation @xmath16 or the _ shifted _ wave equation @xmath17 in @xcite pierfelice derived strichartz estimates for the wave equation ( [ nonshiftedwave ] ) with radial data on a class of riemannian manifolds containing all hyperbolic spaces . the wave equation ( [ nonshiftedwave ] )
was also investigated on the 3dimensional hyperbolic space by metcalfe and taylor @xcite , who proved dispersive and strichartz estimates with applications to small data global well
posedness for the semilinear wave equation .
this result was recently generalized by anker and pierfelice @xcite to other dimensions .
another recent work @xcite by hassani contains a first study of on general riemannian symmetric spaces of noncompact type . to our knowledge ,
the semilinear wave equation ( [ shiftedwave ] ) was first considered by fontaine @xcite in dimension @xmath18 and @xmath19 .
the most famous work involving ( [ shiftedwave ] ) is due to tataru . in @xcite
he obtained dispersive estimates for the operators @xmath20 and @xmath21 acting on inhomogeneous sobolev spaces and then transferred them from @xmath11 to @xmath0 in order to get well posedness results for the euclidean semilinear wave equation ( see also @xcite ) .
though tataru proved dispersive estimates with exponential decay in time , these are not sufficient to obtain actual strichartz estimates on hyperbolic spaces .
complementary results were obtained by ionescu @xcite , who investigated @xmath22 sobolev estimates for the above operators on all hyperbolic spaces . in this paper
we pursue our study of dispersive equations on hyperbolic spaces , initiated with the schrdinger equation @xcite , by considering the shifted wave equation ( [ shiftedwave ] ) on .25mm@xmath11 .
we obtain a wider range of strichartz estimates than in the euclidean setting and deduce stronger well - posedness results .
more precisely , in section [ kernel ] we use spherical harmonic analysis on hyperbolic spaces to estimate the kernel of the operator @xmath23 , where @xmath24 , @xmath25 with @xmath26 , and @xmath27 , @xmath28 are suitable exponents . in section [ dispersive ]
we first deduce dispersive @xmath29 estimates for @xmath30 , when @xmath31.25 mm , by using interpolation and the kunze
stein phenomenon @xcite . in section [ strichartz ]
we next deduce the following strong strichartz estimates for solutions to the cauchy problem ( [ shiftedwave ] ) : @xmath32 where @xmath33 is any ( possibly unbounded ) interval in [email protected] mm , .25mm@xmath35 are admissible couples such that separately @xmath36 and [email protected] mm , @xmath38.25 mm .
notice that the sobolev spaces involved in ( [ strichartzest ] ) are naturally related to the conservation laws of the shifted wave equation ( see section [ sobolev ] ) .
we conclude in section [ lwp ] with an application of to local well posedness of the nonlinear wave equation for initial data with low regularity .
while we obtain the same regularity curve as in the euclidean case for subconformal powerlike nonlinearities , we prove local well
posedness for superconformal powers under lower regularity assumptions on the inital data .
in order to keep down the length of this paper , we postpone applications of the strichartz estimates to global well posedness of the nonlinear wave equation and generalizations of the previous results to damek ricci spaces .
in this paper , we consider the simplest class of riemannian symmetric spaces of the noncompact type , namely real hyperbolic spaces @xmath11 of dimension .25mm@xmath39 .25mm(we shall restrict to .25mm@xmath40 .25mmin section [ lwp]).25 mm .
we refer to helgason s books @xcite and to koornwinder s survey @xcite for their algebraic structure and geometric properties , as well as for harmonic analysis on these spaces , and we shall be content with the following information . @xmath11 can be realized as the symmetric space @xmath41 , where @xmath42 and @xmath43.25 mm . in geodesic polar coordinates on @xmath11 , the riemannian volume writes @xmath44 and the laplace
beltrami operator @xmath45 the spherical functions .25mm@xmath46
on .25mm@xmath11 are normalized radial eigenfunctions of @xmath47 : @xmath48 where .25mm@xmath49
.25mmand [email protected] mm . they can be expressed in terms of special functions : @xmath51 where @xmath52 denotes the jacobi functions and @xmath53 the gauss hypergeometric function . in the sequel
we shall use the integral representations @xmath54 which imply in particular that @xmath55 we shall also use the harish chandra expansion @xmath56 where the harish
chandra @xmath57function is given by @xmath58 and @xmath59 it is well known that there exist @xmath60.25 mm , @xmath61 and @xmath62 such that , for every @xmath63 and @xmath64 with @xmath65.25 mm , @xmath66 we need to improve upon this estimate , by enlarging the domain , by estimating the derivatives of @xmath67 and by gaining some additional decay in @xmath68 .25mmfor @xmath69 .
the following recurrence formula holds : @xmath70 let @xmath71 and @xmath72.25 mm .
then , for every [email protected] mm , there exists .25mm@xmath74 such that @xmath75 consider first the case [email protected] mm .
there exists @xmath77 such that @xmath78.25 mm
. choose @xmath79 such that @xmath80 and @xmath62 such that @xmath81 .
for [email protected] mm , we have [email protected] mm , hence @xmath84 as required . for [email protected] mm
, we have @xmath86 hence @xmath87 derivatives are estimated by the cauchy formula . under suitable assumptions ,
the spherical fourier transform of a bi@xmath88invariant function @xmath89 on @xmath90 is defined by @xmath91 and the following formulae hold : * inversion formula : @xmath92 * plancherel formula : @xmath93 in the sequel we shall use the fact that [email protected] mm , where @xmath95 denotes the abel transform and @xmath96 the fourier transform on the real line . actually we shall use the factorization .25mm@xmath97 .
recall the following expression of the inverse abel transform : @xmath98 if @xmath99 i d odd , the right hand side involves a plain differential operator while , if @xmath99 is even , the fractional derivative must be interpreted as follows : @xmath100
let us first introduce inhomogeneous sobolev spaces on hyperbolic spaces @xmath11 , which will be involved in the conservation laws , in the dispersive estimates and in the strichartz estimates for the shifted wave equation .
we refer to @xcite for more details about functions spaces on riemannian manifolds .
let @xmath101 and @xmath102.25 mm . by definition ,
@xmath103 is the image of @xmath104 under @xmath105 ( in the space of distributions on @xmath11 ) , equipped with the norm @xmath106 in this definition , we may replace @xmath107 by @xmath108 , where [email protected] mm . for simpli- city ,
we choose .25mm@xmath110 .25mmindependently of .25mm@xmath111 .25mmand we set @xmath112 thus @xmath113 and @xmath114 . if .25mm@xmath115
.25mmis a nonnegative integer , then @xmath103 concides with the sobolev space @xmath116 defined in terms of covariant derivatives and equipped with the norm @xmath117 [ set ] let @xmath118 .25mmand @xmath119 such that @xmath120 ( .25mm.]).25 mm .
then @xmath121 by this inclusion , we mean that there exists a constant @xmath62 .25mmsuch that @xmath122 we sketch two proofs . the first one is based on the localization principle for lizorkin triebel spaces @xcite and on the corresponding result in @xmath123 . more precisely , given a tame partition of unity @xmath124 on @xmath11 , we have @xmath125 using the inclusions .25mm@xmath126 .25mmand @xmath127.25 mm , we conclude that @xmath128 the second proof is based on the .25mm@xmath129 mapping properties of the convolution operator .25mm@xmath130 ( see @xcite and the references cited therein ) . beside the @xmath131 sobolev spaces
@xmath132 , our analysis of the shifted wave equation on @xmath11 involves the following @xmath133 sobolev spaces : @xmath134 where [email protected] mm , .25mm@xmath102 .25mmand
.25mm@xmath136 .25mm(actually we are only interested in the cases .25mm@xmath137 .25mmand [email protected] mm ) .
notice that @xmath139 if @xmath140.25 mm , then @xmath141 where @xmath142 ( recall that @xmath143 is decreasing as @xmath144 and @xmath145 .
let [email protected] mm .
we have [email protected] mm ,
.25mmwhere @xmath148 by the inversion formula for the spherical fourier transform on .25mm@xmath11 .
let us split up the integral @xmath149 and the kernel @xmath150 accordingly .
on the one hand , @xmath151 maps @xmath13 into @xmath152 . on the other hand , @xmath153 is a radial kernel in @xmath154 , hence @xmath155{\hskip.25mm}}(d)\,\widetilde
d^{-\sigma}{\hskip.25mm}d^{-\tau}f = f{\hskip-.25mm}*{\hskip-.25mm}k_{{\hskip.25mm}\sigma,\tau}^{\,0}\ ] ] maps @xmath13 into @xmath156 by the kunze - stein phenomenon .
thus @xmath157 belongs to @xmath158 , as required .
let us next introduce the energy @xmath159 for solutions to the homogeneous cauchy problem @xmath160 it is easily verified that @xmath161.25 mm , hence is conserved . in other words , for every time .25mm@xmath162
.25mmin the interval of definition of [email protected] mm , @xmath164 let .25mm@xmath102 .25mmand [email protected] mm . by applying the operator .25mm@xmath165 to ( [ wavehom ] )
, we deduce that @xmath166 which can be rewritten in terms of sobolev norms as follows : @xmath167
in this section we derive pointwise estimates for the radial convolution kernel .25mm@xmath168 of the operator .25mm@xmath169 , for suitable exponents @xmath102 and @xmath170.25 mm . by the inversion formula of the spherical fourier transform , @xmath171 contrarily to the euclidean case
, this kernel has different behaviors , depending whether .25mm@xmath162 .25mmis small or large , and therefore we can not use any rescaling .
let us split up @xmath172 using smooth cut off functions @xmath173 and @xmath174 on @xmath175 such that @xmath176.25 mm , @xmath177 on @xmath178 $ ] and @xmath179 on @xmath180.25 mm .
we shall first estimate .25mm@xmath181 and next a variant of .25mm@xmath182 .
the kernel .25mm@xmath182 has indeed a logarithmic singularity on the sphere .25mm@xmath183
.25mmwhen .25mm@xmath184 .
we bypass this problem by considering the analytic family of operators @xmath185 in the vertical strip .25mm@xmath186
.25mmand the corresponding kernels @xmath187 notice that the gamma function , which occurs naturally in the theory of riesz distributions , will allow us to deal with the boundary point .25mm@xmath188 , while the exponential function yields boundedness at infinity in the vertical strip .
notice also that , once multiplied by [email protected] mm , the operator .25mm@xmath190
.25mmbehaves like [email protected] mm . [ estimatewt0 ]
let @xmath102 and @xmath193.25 mm .
the following pointwise estimates hold for the kernel @xmath194 * assume that @xmath195.25 mm . then , for every @xmath196.25 mm , @xmath197 * assume that @xmath198.25 mm . * * if @xmath199.25 mm , then @xmath200 * * if @xmath201.25 mm , then @xmath202 recall that @xmath203 by symmetry we may assume that [email protected] mm .
\(i ) it follows from the estimates and that @xmath205 \(ii ) we prove first ( a ) by substituting in the first integral representation of @xmath46 in and by reducing this way to fourier analysis on [email protected] mm .
specifically , @xmath206 where [email protected] mm , up to a positive constant . according to the estimate ( [ estimatec ] ) and to lemma a.1 in appendix a , the inner integral is bounded above by @xmath208 since @xmath209 we conclude that @xmath210 we prove next ( b ) by substituting in the harish
chandra expansion ( [ hcexpansion ] ) of @xmath46 and by reducing again to fourier analysis on [email protected] mm . specifically , @xmath211 where @xmath212 and @xmath213 by applying lemma a.1 and by using the estimates for @xmath67 and its derivatives , we obtain @xmath214 and @xmath215 we conclude the proof by summing up these estimates in .
[ estimatewtildetinfty ] the following pointwise estimates hold for the kernel @xmath216 , for any fixed .25mm@xmath217 and uniformy in .25mm@xmath218 with .25mm@xmath219 * assume that @xmath220.25 mm .
* * if @xmath221.25 mm , then @xmath222 * * if @xmath223.25 mm , then @xmath224.25 mm . *
assume that @xmath225.25 mm
. then @xmath226 _ proof of theorem [ estimatewtildetinfty].ii .
_ recall that , up to a positive constant , @xmath227 by symmetry we may assume again that [email protected] mm . if [email protected] mm , we resume the proof of theorem [ estimatewt0].ii.a , using lemma a.2 instead of lemma a.1 , and estimate this way @xmath229 if [email protected] mm , we resume the proof of theorem [ estimatewt0].ii.b and expand this way @xmath231 where @xmath232 and @xmath233 it follows from the expression of the @xmath234function and from the estimates of the coefficients @xmath67 .25mmthat
@xmath235 .25mmis a symbol of order @xmath236 by applying lemma a.2 , we obtain the following estimates of the expressions @xmath237 , except for @xmath238 : @xmath239 , @xmath240.25 mm , @xmath241 as far as .25mm@xmath238 .25mmis concerned , lemma a.2 yields the estimates @xmath242 the second one can be improved by applying lemma a.3 instead of lemma a.2 . for this purpose ,
let us establish the asymptotic behavior of the symbol [email protected] mm , as .25mm@xmath244 .
on the one hand , @xmath245 according to stirling s formula @xmath246 on the other hand , @xmath247 hence @xmath248 .25mmand [email protected] mm , we obtain @xmath250 instead of the singularity .25mm@xmath251
.25mmin , the estimate of .25mm@xmath238
.25mminvolves this time the singularity [email protected] mm , which cancels with the denominator of the front expression @xmath253 in .
notice moreover that the numerator of yields enough decay to get uniform bounds in [email protected] mm . in conclusion , by combining , , , , , we obtain @xmath254 @xmath255 [ estimatewtinfty ] the kernel @xmath256 can be estimated in the same way , except that @xmath257 when @xmath258 is close to @xmath259.25 mm .
let us turn to the small time estimates in theorem [ estimatewtildetinfty ] .
the estimate ( i.a ) is of local nature and thus similar to the euclidean case . for the sake of completeness
, we include a proof in appendix c. it remains for us to prove the estimate ( i.b ) .
_ proof of theorem [ estimatewtildetinfty].i.b .
_ here .25mm@xmath260
.25mmand [email protected] mm . by symmetry
we may assume again that [email protected] mm .
we use now the inverse abel transform given by formulae and .
up to positive constants , the inverse spherical fourier transform can be rewritten in the following way : @xmath261 where @xmath262 let us split up @xmath263 and @xmath264 .25mmaccordingly , so that @xmath265 _ case 1 _ : assume that .25mm@xmath266
.25mmis odd .
first of all , let us expand @xmath267 since the coefficients .25mm@xmath268 are linear combinations of products @xmath269 with @xmath270 , and .25mm@xmath271
.25mmis [email protected] mm , as well as its deri- vatives , we deduce that .25mm@xmath268 .25mmis .25mm@xmath273
.25mmas @xmath274.25 mm .
consider next @xmath275 according to lemma a.2 , for every @xmath276-.25 mm , there exists .25mm@xmath277 .25mmsuch that @xmath278 as a conclusion , @xmath279 _ case 2 _ : assume that .25mm@xmath280 .25mmis even .
according to case 1.25 mm , for every @xmath281 , there exists .25mm@xmath282 .25mmsuch that @xmath283 by estimating @xmath284 and performing the change of variables [email protected] mm , we deduce that @xmath286 @xmath255
in this section we obtain @xmath287 estimates for the operator @xmath288 , which will be crucial for our strichartz estimates in next section .
let us split up its kernel .25mm@xmath289 as before .
we will handle the contribution of .25mm@xmath290 , using the pointwise estimates obtained in subsection [ kernelestimatewt0 ] and the following criterion based on the kunze - stein phenomenon .
[ ks ] there exists a constant @xmath62 such that , for every radial measurable function @xmath291 on @xmath11 , for every @xmath292 and .25mm@xmath293 , @xmath294 where @xmath295 and @xmath296 .25 mm ( ) .
this estimate is obtained by complex multilinear interpolation between the following version @xcite of the kunze
stein phenomenon @xmath297 and the elementary inequalities @xmath298 specifically , if .25mm@xmath31 .25mmand [email protected] mm , define .25mm@xmath300
.25mmby [email protected] mm . by complex interpolation ,
let us deduce the intermediate estimate @xmath302 from the endpoint estimates @xmath303 and @xmath304 here @xmath305 are linear combinations with nonzero complex coefficients of characteristic functions of disjoints borel sets in .25mm@xmath11 with finite positive measure , the @xmath306 s being moreover spherical . as in the proof of the riesz
thorin theorem ( see for instance @xcite ) , we assume that [email protected] mm , we consider the analytic families of simple functions @xmath308 and we apply the hadamard three lines theorem to the holomorphic function @xmath309 in the vertical strip [email protected] mm . more precisely , if [email protected] mm , then @xmath312 hence [email protected] mm , according to . similarly , if [email protected] mm , then @xmath315 hence .25mm@xmath316 , according to . the estimate is obtained by applying the three lines theorem to .25mm@xmath317 .25mmat the point [email protected] mm , where @xmath319 eventually , the symmetric case , where .25mm@xmath320 .25mmand [email protected] mm , .25mmis handled similarly . for the second part .25mm@xmath322 , we resume the euclidean approach , which consists in interpolating analytically between @xmath323 and @xmath324 estimates for the family of operators @xmath325 in the vertical strip [email protected] mm .
[ dispersive0 ] assume that @xmath327.25 mm , [email protected] mm , .25mm@xmath329 and @xmath330.25 mm .
@xmath331 we divide the proof into two parts , corresponding to the kernel decomposition .25mm@xmath289 . by applying lemma [ ks ] and by using the pointwise estimates in theorem [ estimatewt0].i
, we obtain on one hand @xmath332 for the second part , we consider the analytic family . if [email protected] mm , then @xmath334 if .25mm@xmath335 , we deduce from the pointwise estimates in theorem [ estimatewtildetinfty].i that @xmath336 by interpolation we conclude for .25mm@xmath337 .25mmthat @xmath338 [ dispersiveinfty ] assume that @xmath339.25 mm , [email protected] mm , .25mm@xmath329 and @xmath330.25 mm .
then @xmath340 we divide the proof into three parts , corresponding to the kernel decomposition @xmath341 _ estimate 1 _ : by applying lemma [ ks ] and using the pointwise estimates in theorem [ estimatewt0].ii.a , we obtain @xmath342 _ estimate 2 _ : by applying lemma [ ks ] and using the pointwise estimates in theorem [ estimatewt0].ii.b , we obtain @xmath343 _ estimate 3 _ : in order to estimate the .25mm@xmath344 norm of .25mm@xmath345 , we may apply lemma [ ks ] and use pointwise estimates of .25mm@xmath346 ( see remark [ estimatewtinfty ] ) . while @xmath347 are .25mm@xmath348 .25mmfor any [email protected] mm , the integral @xmath349 is finite provided [email protected] mm , which is too large compared with the critical exponent [email protected] mm .
instead we use again interpolation for the analytic family . if .25mm@xmath352 , then @xmath353 if .25mm@xmath335 , we deduce from theorem [ estimatewtildetinfty].ii that @xmath354 by interpolation we conclude for .25mm@xmath355 that @xmath356 by taking .25mm@xmath357
.25mmin theorems [ dispersive0 ] and [ dispersiveinfty ] , we obtain in particular the following dispersive estimates .
[ dispersiveglobal ] let @xmath31 and @xmath358 .
then @xmath359 with @xmath360 replaced by @xmath361 in dimension @xmath19.25 mm .
notice that tataru .25mm@xcite obtained dispersive estimates with exponential decay in time for the operators @xmath362 and @xmath363.25 mm , but did not prove actual strichartz estimates . here
we obtain dispersive estimates with polynomial decay in time for the operator @xmath364.25 mm .
this difference reflects the fact that the fourier multipliers associated with the operators @xmath362 and @xmath363 are analytic in a strip of the complex plane , which is not the case of @xmath364.25 mm . by applying lemma [ ks ] in full generality , we obtain the following decoupled estimate for the operators @xmath365 [ decoupledwt0 ]
let @xmath366.25 mm , .25mm@xmath367 and [email protected] mm
. then @xmath368
we shall assume .25mm@xmath40
.25mmthroughout this section and discuss the 2dimensional case in the final remark . consider the inhomogeneous linear wave equation on .25mm@xmath11 : @xmath369 whose solution is given by duhamel s formula : @xmath370 [ admcouple ] a couple @xmath371 is called admissible if @xmath372 belongs to the triangle @xmath373\!\times\!\bigl(0,\frac12\bigr ) \bigm|\frac2p{\hskip-.25mm}+{\hskip-.25mm}\frac{n-1}q{\hskip-.25mm}\ge{\hskip-.25mm}\frac{n-1}2\,\bigr\}\ ] ] ( see figure 1 ) .
[ c]@xmath374 [ c]@xmath375 [ c]@xmath376 [ c]@xmath377 [ c]@xmath378 [ c]@xmath379 , title="fig:",width=264 ] [ dim3end ] observe that the endpoint @xmath380 is included in the triangle @xmath381 in dimension @xmath382 but not in dimension @xmath383 ( see figure 2 ) .
[ c]@xmath374 [ c]@xmath375 [ c]@xmath376 [ c]@xmath378 [ c]@xmath379 , title="fig:",width=264 ] [ strichartzestimates ] let @xmath371 and @xmath384 be two admissible couples .
then the following strichartz estimate holds for solutions to the cauchy problem : @xmath385 where @xmath386 and @xmath387 .
moreover , @xmath388 consider the operator @xmath389 initially defined from @xmath13 into @xmath390 , and its formal adjoint @xmath391 initially defined from @xmath392 into @xmath393 .
the .25mm@xmath394 -.25mmmethod consists in proving first the .25mm@xmath395 .25mmboundedness of the operator @xmath396 and of its truncated version @xmath397 for every admissible couple @xmath371 and for every @xmath398 , and in decoupling next the indices .
assume that the admissible couple .25mm@xmath371 .25mmis different from the endpoint [email protected] mm .
then we deduce from corollary [ dispersiveglobal ] that the norms @xmath400 and @xmath401 are bounded above by @xmath402 where @xmath403 . on the one hand ,
the convolution kernel .25mm@xmath404 defines obviously a bounded operator from .25mm@xmath405 to .25mm@xmath406 , for all [email protected] mm , in particular from .25mm@xmath408 to .25mm@xmath409 , since [email protected] mm . on the other hand , the convolution kernel @xmath411 .25mmwith
.25mm@xmath412 .25mmdefines a bounded operator from .25mm@xmath405 to .25mm@xmath413 , for all .25mm@xmath414 .25mmsuch that .25mm@xmath415 , .25mmin particular from .25mm@xmath408 to .25mm@xmath409 , since .25mm@xmath410 .25mmand [email protected] mm .
at the endpoint [email protected] mm , we have [email protected] mm .
thus the previous argument breaks down and is replaced by the refined analysis carried out in @xcite .
notice that the problem lies only in the first part of and not in the second one , which involves an integrable convolution kernel on [email protected] mm .
thus @xmath394 and @xmath419 are bounded from .25mm@xmath420 to [email protected] mm , for every admissible couple @xmath371.25 mm . as a consequence , @xmath422
is bounded from .25mm@xmath420 .25mmto .25mm@xmath13 .25mmand
@xmath423 .25mmis bounded from .25mm@xmath13 .25mmto .25mm@xmath421 .
in particular , @xmath424 and @xmath425 we next decouple the indices .
let @xmath426 be two admissible couples and let @xmath398 , @xmath427 .
since @xmath423 and @xmath422 are separately continuous , the operator @xmath428 is bounded from @xmath429 to @xmath421.25 mm .
according to @xcite , this result remains true for the truncated operator @xmath430 and hence for @xmath431 as long as .25mm@xmath432
.25mmand .25mm@xmath433 .25mmare not both equal to [email protected] mm .
we handle the remaining case , where .25mm@xmath435 .25mmand [email protected] mm , by combining the bilinear approach in @xcite with our previous estimates .
specifically let us split up again .25mm@xmath437 , using smooth cut off functions , and .25mm@xmath438 accordingly .
on one hand , it follows from proposition [ decoupledwt0 ] that @xmath439 is bounded from .25mm@xmath429 .25mmto [email protected] mm , .25mmfor every .25mm@xmath440 .25mmand [email protected] mm , in particular for .25mm@xmath435 .25mmand [email protected] mm .
as far as it is concerned , the @xmath443 boundedness of @xmath444 amounts to estimating the hermitian form @xmath445 by .25mm@xmath446 .
let us split up dyadically @xmath447 and @xmath448 .25mmaccordingly .
for every [email protected] mm , let us further split up @xmath450 notice the orthogonality @xmath451 and the almost orthogonality @xmath452 we claim that @xmath453 when .25mm@xmath454
.25mmand [email protected] mm . these estimates will be obtained by complex interpolation between the following cases ( see figure 3 ) : * .25mm@xmath456 .25mmand [email protected] mm , * .25mm@xmath458 .25mmand [email protected] mm , * [email protected] mm .
[ interpolation ] [ c]@xmath374 [ c]@xmath376 [ c]@xmath377 [ c]@xmath379 [ c]@xmath461 _ case _ ( _ _ a__.25 mm ) : assume that [email protected] mm , @xmath457 .25mmand [email protected] mm , [email protected] mm .
consider the operators @xmath463 and @xmath464 by resuming the proof of theorem [ dispersive0 ] and by applying the .25mm@xmath465 argument , we obtain that .25mm@xmath466 is bounded from .25mm@xmath467 .25mmto [email protected] mm , where [email protected] mm . by combining this result with hlder s inequality , we deduce that @xmath469 with [email protected] mm . _ case _ ( _ _ b__.25 mm ) : if [email protected] mm , @xmath459 .25mmand [email protected] mm , @xmath473.25 mm , we have symmetrically @xmath474 _ case _ ( _ _ c__.25 mm ) : assume that .25mm@xmath460
.25mmand [email protected] mm .
let us rewrite @xmath476 by using the dispersive estimates @xmath477 ( see the proofs of theorems [ dispersive0 ] and [ dispersiveinfty ] ) , we obtain @xmath478 hence , by hlder s inequality , @xmath479 our claim follows now by complex interpolation between the estimates obtained in cases ( a ) , ( b ) and ( c ) above . by summing up and by using hlder s inequality ,
we conclude that @xmath480 @xmath481 @xmath482 if [email protected] mm .
notice that .25mm@xmath484
.25mmunder this assumption .
let us turn to .
on the one hand , the energy estimate ( [ conservationenergy ] ) yields @xmath485 for every [email protected] mm . on the other hand , since @xmath422 is bounded from @xmath429 to @xmath13 , both expressions @xmath487 and @xmath488 are bounded above by @xmath489 uniformly in [email protected] mm .
we conclude the proof of by summing up the previous estimates and by taking the supremum over .25mm@xmath490 .
[ strichartzi ] observe that , in the statement of theorem [ strichartzestimates ] , we may replace @xmath34 by any time interval @xmath33 containing @xmath374.25 mm .
[ dim2end ] an analogous result holds in dimension @xmath19 and its proof is similar , except for the first convolution kernel in , which becomes @xmath491 with @xmath492 and @xmath493.25 mm .
it turns out that , in this case , a couple @xmath371 is admissible if @xmath372 belongs to the region @xmath494\!\times\!(0,\frac12 ) \bigm|\frac2p\!+\!\frac1q\!>\!\frac12\,\bigr\}$ ] ( see figure 4 ) .
[ c]@xmath374 [ c]@xmath375 [ c]@xmath376 [ c]@xmath495 [ c]@xmath378 [ c]@xmath379 , title="fig:",width=264 ]
we shall assume .25mm@xmath497
.25mmthroughout this section and discuss the lower dimensional cases .25mm@xmath18 .25mmand
.25mm@xmath19 .25mmin the final remarks .
we apply strichartz estimates for the inhomogeneous linear cauchy problem associated with the wave equation to prove local well posedness results for the following nonlinear cauchy problem @xmath498 with a power like nonlinearity .25mm@xmath499 . by this
we mean that @xmath500 .25mmand [email protected] mm . let us recall the definition of local well posedness .
the nlw cauchy problem is .25mmlocally well posed in .25mm@xmath502 if , for any bounded subset .25mm@xmath503 of .25mm@xmath502 , there exist .25mm@xmath504 and a banach space .25mm@xmath505 , continuously embedded into .25mm@xmath506{\hskip.25mm};h^{s,\tau})\cap c^1({\hskip.25mm}[-t , t{\hskip.25mm}]{\hskip.25mm};h^{s,\tau-1})$].25 mm , such that @xmath507 for any initial data .25mm@xmath508 , @xmath509 has a unique solution .25mm@xmath510 , @xmath507 the map .25mm@xmath511 is continuous from .25mm@xmath503 into .25mm@xmath505 .
the amount of smoothness @xmath27 requested for lwp of in @xmath512 depends on @xmath513 and is represented in figure 5 .
[ lwpn ] [ c]@xmath27 [ c]@xmath374 [ c]@xmath375 [ c]@xmath376 [ c]@xmath514 [ c]@xmath513 [ c]@xmath515 [ c]@xmath516 [ c]@xmath517 [ c]@xmath518 [ c]@xmath519 [ c]@xmath520 [ c]@xmath521 , title="fig:",width=377 ] there @xmath522 @xmath523 @xmath524 and the curves .25mm@xmath519 ,
@xmath520 , @xmath521 are given by @xmath525 when [email protected] mm , we obtain the same regularity curve as in the euclidean case .
since our strichartz estimates hold for a large family of admissible pairs , they are sufficient to study the regularity problem via a fixed point argument ; in the euclidean setting this problem was solved by different methods , depending on the range of the power @xmath513 involved in the nonlinearity and on the regularity of initial data .
[ wpl2 ] let .25mm@xmath497 and assume that .25mm@xmath499 satisfies .
then the nlw is locally well posed in .25mm@xmath512 in the following cases@xmath527 * @xmath528 and @xmath529 * @xmath530 and @xmath531 * @xmath532 and @xmath533 * @xmath534 and @xmath535 more precisely , for all such nonlinearity power .25mm@xmath513 and regularity .25mm@xmath27 , there exists a positive @xmath423 , depending on the initial data , and a unique solution .25mm@xmath163 to nlw such that @xmath536{\hskip.25mm};h^{\sigma-\frac12,\frac12}({\mathbb{h}^n})\bigr)\ , \cap{\hskip.25mm}l^{p_0}\bigl({\hskip.25mm}[-t , t{\hskip.25mm}]{\hskip.25mm};l^{q_0}({\mathbb{h}^n}))\,,\ ] ] for a suitable admissible couple @xmath537 , and @xmath538{\hskip.25mm};h^{\sigma-\frac12,-\frac12}({\mathbb{h}^n})\bigr)\,.\ ] ] we apply the standard fixed point method based on strichartz estimates .
define @xmath539 as the solution to the following linear cauchy problem @xmath540 which is given by the duhamel formula @xmath541 we deduce from the strichartz estimates , and from remark [ strichartzi ] that @xmath542{\hskip.25mm};{\hskip.25mm}h^{\sigma-\frac12,\frac12}\big ) } + \|\partial_tu\|_{l^{\infty}\big([-t , t{\hskip.25mm}]{\hskip.25 mm } ; { \hskip.25mm}h^{\sigma-\frac12,-\frac12}\big ) } + \|u\|_{l^p([-t , t{\hskip.25mm}];{\hskip.25mm}l^q\vphantom{l_t^{\tilde p'}})}\\ & \lesssim\,\|f\|_{h^{\sigma-\frac12,\frac12 } } + \,\|g\|_{h^{\sigma-\frac12,-\frac12 } } + \,\|f(v)\|_{l^{\tilde{p}'}{\hskip-.25mm}\bigl([-t,{\hskip.25mm}t{\hskip.25mm}]{\hskip.25 mm } ; { \hskip.25mm}h^{{\sigma+\tilde{\sigma}}-1}_{\tilde{q}'}\bigr)}\ , , \end{aligned}\ ] ] for all admissible couples @xmath371 , @xmath543 introduced in definition [ admcouple ] , for all .25mm@xmath398 , @xmath427 , and for a positive .25mm@xmath423 to be determined later . according to the nonlinear assumption , we estimate the inhomogeneous term as follows : @xmath544{\hskip.25 mm } ; { \hskip.25mm}\,h^{{\sigma+\tilde{\sigma}}-1}_{\tilde q'}\big ) } \lesssim \,\|\,|v|^\gamma\|_{l^{\tilde p'}\big([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}h^{{\sigma+\tilde{\sigma}}-1}_{\tilde q'}\big)}{\hskip.25mm}.\ ] ] assuming [email protected] mm , we deduce from sobolev s embedding ( proposition [ set ] ) that @xmath546{\hskip.25mm};{\hskip.25mm}h^{\sigma-\frac12,\frac12}\big ) } + \,\|\partial_tu\|_{l^{\infty}\big([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}h^{\sigma-\frac12,-\frac12}\big ) } + \,\|u\|_{l^p([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^q\vphantom{l_t^{\tilde p'}})}\\ & \lesssim\,\|f\|_{h^{\sigma-\frac12,\frac12 } } + \,\|g\|_{h^{\sigma-\frac12,-\frac12 } } + \,\|v\|_{l^{\tilde{p}'\gamma}\big([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^{\tilde{q}_1'{\hskip-.25mm}\gamma}\big)}^{\,\gamma}{\hskip.25mm}. \end{aligned}\ ] ] in order to remain within the same function space , we require that [email protected] mm . after applying hlder s inequality in time ,
we obtain @xmath548;\,h^{\sigma-\frac12,\frac12}\big ) } + \,\|\partial_tu\|_{l^{\infty}\big([-t , t];\,h^{\sigma-\frac12,-\frac12}\big ) } + \,\|u\|_{l^p([-t , t];\,l^q)\vphantom{h^{\frac12}}}\\ & \lesssim\,\|f\|_{h^{\sigma-\frac12,\frac12}}+ \,\|g\|_{h^{\sigma-\frac12,-\frac12 } } + \,t^{{\hskip.25mm}\lambda}\,\|v\|_{l^p([-t , t];\,l^q)\vphantom{h^{\frac12}}}^{\,\gamma}\ , .
\end{aligned}\ ] ] here we have assumed .25mm@xmath549 .25mmand set [email protected] mm .
it remains for us to check that the following conditions can be fulfilled simultaneously : @xmath551{\hskip-.25mm}\times{\hskip-.25mm}\bigl(\frac{n-3}{2{\hskip.25mm}(n-1)},\frac12\bigr)\,,\\ \;\text{(vii ) } & \bigl(\frac1{\tilde{p}},\frac1{\tilde{q}}\bigr){\hskip-.25mm}\in{\hskip-.25mm}\bigl(0,\frac12\bigr]{\hskip-.25mm}\times{\hskip-.25mm}\bigl(\frac{n-3}{2{\hskip.25mm}(n-1)},\frac12\bigr)\ , .
\end{cases}\ ] ] suppose indeed that there exist indices .25mm@xmath552 .25mmsatisfying all conditions in ( [ condts ] )
. then shows that .25mm@xmath553 .25mmmaps
.25mm@xmath554 .25mminto itself , where .25mm@xmath554 .25mmdenotes the banach space @xmath555{\hskip.25mm};h^{\sigma-\frac12,\frac12}({\mathbb{h}^n } ) ) { \hskip.25mm}\cap{\hskip.25mm}l^p({\hskip.25mm}[-t , t{\hskip.25mm}]{\hskip.25mm};l^q({\mathbb{h}^n}))\,,\\ & \partial_{{\hskip.25mm}t{\hskip.25mm}}u{\hskip-.25mm}\in{\hskip-.25mm}c{\hskip.25mm}([-t , t{\hskip.25mm}]{\hskip.25mm};h^{\sigma-\frac12,-\frac12}({\mathbb{h}^n}))\,\bigr\}\ , , \end{aligned}\ ] ] equipped with the norm @xmath556{\hskip.25mm};{\hskip.25mm}h^{\sigma-\frac12,\frac12}\bigr ) } + \,\|\partial_{{\hskip.25mm}t{\hskip.25mm}}u\|_{l^\infty\bigl({\hskip.25mm}[-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}h^{\sigma-\frac12,-\frac12}\bigr ) } + \,\|u\|_{l^p\bigl({\hskip.25mm}[-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^q\bigr)}\,,\ ] ] moreover we shall show that @xmath553 is a contraction on the ball @xmath557 provided the time .25mm@xmath504 .25mmis sufficiently small and the radius .25mm@xmath558 .25mmis sufficiently large .
.25mmand [email protected] mm , @xmath560.25 mm . by arguing as above and using hlder s inequality , we have @xmath561{\hskip.25mm};{\hskip.25mm}h_{\tilde{q}'}^{\sigma+\tilde{\sigma}-1}\big)}\\ & \le\,c\;\|{\hskip.25mm}\{{\hskip.25mm}|v|^{\gamma-1}\!+|\tilde v|^{\gamma-1}{\hskip.25mm}\}\ , _ { l^{\tilde{p}'}\bigl({\hskip.25mm}[-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^{\tilde{q}_1'}\bigr)}\\ & \le\,c\;t^{{\hskip.25mm}\lambda}\ , \bigl\{{\hskip.25mm}\|v\|_{l^p\bigl([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^q\bigr)}^{\,\gamma-1}\
! + \|\tilde{v}\|_{l^p\bigl([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^q\bigr)}^{\,\gamma-1}\bigr\}\ , \|{\hskip.25mm}v{\hskip-.25mm}-{\hskip-.25mm}\tilde{v}{\hskip.25mm}\|_{l^p\bigl([-t , t{\hskip.25mm}]{\hskip.25mm};{\hskip.25mm}l^q\bigr)}\\ & \le\,c\;t^{{\hskip.25mm}\lambda}\ , \bigl\{{\hskip.25mm}\|v\|_x^{\gamma-1}\!+\|\tilde{v}\|_x^{\gamma-1}{\hskip.25mm}\bigr\}\ , \|{\hskip.25mm}v{\hskip-.25mm}-{\hskip-.25mm}\tilde{v}{\hskip.25mm}\|_x\ , .
\end{aligned}\ ] ] if .25mm@xmath562 .25mmand [email protected] mm , then yields on the one hand @xmath564 and @xmath565 while yields on the other hand @xmath566 thus , if we choose .25mm@xmath567
.25mmso large that .25mm@xmath568 .25mmand
.25mm@xmath504 .25mmso small that
.25mm@xmath569 .25mmand [email protected] mm , then @xmath571 if .25mm@xmath572 and [email protected] mm , @xmath574.25 mm .
hence the map .25mm@xmath553 .25mmis a contraction on the complete metric space @xmath575 and the fixed point theorem allows us to conclude .
let us eventually prove the existence of couples @xmath371 and @xmath543 satisfying all conditions in .
condition ( [ condts].iii ) amounts to @xmath576 by combining with ( [ condts].ii ) and ( [ condts].vi ) , we deduce that @xmath577 this implies that [email protected] mm . by combining with ( [ condts].vii ) , we obtain @xmath579 by combining ( [ cond0 ] ) with ( [ condts].vii ) , we also obtain [email protected] mm . in summary , the conditions on @xmath111 reduce to @xmath581 or case by case to * @xmath582 .25mmand [email protected] mm , * @xmath584 .25mmand [email protected] mm , * @xmath586 .25mmand [email protected] mm .
let us turn to the indices .25mm@xmath432
.25mmand [email protected] mm . according to
, we have @xmath588 and @xmath589 by taking into account the previous conditions on [email protected] mm , we end up with the following conditions on .25mm@xmath432 .25mmand .25mm@xmath433 : @xmath590 there exist indices .25mm@xmath432 .25mmand
.25mm@xmath433 .25mmwhich satisfy provided that [email protected] mm .
we thus have to find .25mm@xmath592 .25mmsuch that @xmath593 with [email protected] mm .
this implies that .25mm@xmath111 .25mmhas to satisfy the following conditions : @xmath595 with [email protected] mm . the fact that .25mm@xmath597
.25mmeasily implies that [email protected] mm . the fact that .25mm@xmath599 .25mmimplies that [email protected] mm . in summary , here are the final conditions on [email protected] mm , depending on .25mm@xmath513 .25mmand possibly on the dimension .25mm@xmath99 : * @xmath601 .25mmand [email protected] mm . * @xmath603 .25mmand [email protected] mm . *
@xmath605 .25mmand .25mm@xmath606 .25mmwhen [email protected] mm .
+ when [email protected] mm , we distinguish two subcases : * * @xmath609 .25mmand [email protected] mm , * * @xmath610 .25mmand [email protected] mm . * when [email protected] mm , we distinguish two subcases : * * @xmath613 .25mmand [email protected] mm , * * @xmath615 .25mmand [email protected] mm .
+ when [email protected] mm , we replace .25mm@xmath618
.25mmby [email protected] mm .
+ when [email protected] mm , .25mm@xmath620 .25mmand [email protected] mm .
let us now examine these cases separately .
* case ( a ) . * in this case , we choose successively .25mm@xmath111 .25mmsuch that @xmath621 @xmath622 .25mmsatisfying , and [email protected] mm , @xmath433 .25mmsatisfying .
thus , when .25mm@xmath623
.25mmand [email protected] mm , there exists always an admissible couple @xmath371 such that all conditions are satisfied and [email protected] mm .
* case ( b)*. in this case , we choose successively .25mm@xmath111 .25mmsuch that @xmath626 @xmath622 .25mmsatisfying , and [email protected] mm , @xmath433 .25mmsatisfying . and a correspondent @xmath622 which satisfies .
thus , when .25mm@xmath584
.25mmand [email protected] mm , there exists always an admissible couple @xmath371 such that all conditions are satisfied and [email protected] mm .
* case ( c ) .
* assume first that [email protected] mm . we choose successively .25mm@xmath111 .25mmsuch that @xmath628 @xmath622 .25mmsatisfying , and [email protected] mm , @xmath433 .25mmsatisfying .
assume next that [email protected] mm .
if [email protected] mm , we choose .25mm@xmath111 .25mmaccording to . if .25mm@xmath629 , we replace by @xmath630 in both cases , we can choose afterwards .25mm@xmath631 .25mmsatisfying and . in summary , when .25mm@xmath605
.25mmand [email protected] mm , there exists always an admissible couple @xmath371 such that all conditions are satisfied and [email protected] mm .
* case ( d ) . * assume first that [email protected] mm . if [email protected] mm , we choose successively .25mm@xmath111 .25mmsuch that @xmath634 @xmath622 .25mmsatisfying , and [email protected] mm , @xmath433 .25mmsatisfying .
if [email protected] mm , is replaced by @xmath635 assume next that [email protected] mm .
we choose again .25mm@xmath111 .25mmaccording to if .25mm@xmath633 .25mmand according to if [email protected] mm . in both cases
, we can choose afterwards .25mm@xmath631 .25mmsatisfying and .
assume eventually that [email protected] mm .
then we choose .25mm@xmath111 .25mmaccording to and .25mm@xmath631 .25mmsatisfying and ( [ condindecesp ] ) . in summary , when .25mm@xmath637
.25mmand [email protected] mm , there exists always an admissible couple @xmath371 such that all conditions are satisfied and [email protected] mm .
this concludes the proof of theorem [ wpl2 ] .
notice that , in dimension @xmath18.25 mm , the strichartz estimates are available in the triangle @xmath640 without the endpoint ( see remark [ dim3end ] ) . by arguing as above
, we prove that the nlw is locally well posed in .25mm@xmath641 if ( see figure 6 ) * @xmath642 @xmath643 and .25mm@xmath624 ; * @xmath642 @xmath644 and .25mm@xmath645 ; * @xmath642 @xmath646 and [email protected] mm . [ lwp3 ] [ c]@xmath27 [ c]@xmath374 [ c]@xmath375 [ c]@xmath376 [ c]@xmath648 [ c]@xmath513 [ c]@xmath649 [ c]@xmath650 [ c]@xmath651 [ c]@xmath520 [ c]@xmath521 , title="fig:",width=377 ] in dimension @xmath19.25 mm , the strichartz estimates are available in the region @xmath652 ( see remark [ dim2end ] ) . by following again the same line of the above proof
, we obtain that the nlw is locally well posed in .25mm@xmath641 if ( see figure 7 ) * @xmath642 @xmath653 and .25mm@xmath654 ; * @xmath642 @xmath655 and .25mm@xmath656 ; * @xmath642 @xmath657 and .25mm@xmath658 ; * @xmath642 @xmath659 and .25mm@xmath660 .
[ c]@xmath27 [ c]@xmath374 [ c]@xmath375 [ c]@xmath495 [ c]@xmath376 [ c]@xmath434 [ c]@xmath661 [ c]@xmath513 [ c]@xmath662 [ c]@xmath663 [ c]@xmath664 [ c]@xmath520 [ c]@xmath521 , title="fig:",width=453 ]
in this appendix we collect some lemmata in fourier analysis on @xmath34 which are used for the kernel analysis in section [ kernel ] and in appendix c. * lemma a.1.*[lemmaa1 ] _ let .25mm@xmath665 be a compactly supported homogeneous symbol on .25mm@xmath34 of order .25mm@xmath666 . in other words , .25mm@xmath665 is a smooth function on @xmath667 , whose support is bounded in .25mm@xmath34 and which has the following behavior at the origin : @xmath668 then its fourier transform @xmath669 is a smooth function on @xmath34.25 mm , with the following behavior at infinity : @xmath670 more precisely , let .25mm@xmath671 be the smallest integer @xmath672.25 mm . then @xmath673.25 mm , @xmath674 , @xmath675 _
let us split up @xmath676 and @xmath677 accordingly , using a homogeneous dyadic partition of unity @xmath678 on [email protected] mm .
notice that .25mm@xmath680 hence .25mm@xmath681 vanishes for @xmath682 large , since @xmath665 is compactly supported . by the leibniz formula
, we obtain , for every [email protected] mm , @xmath684 let .25mm@xmath685 such that [email protected] mm . then @xmath687 * lemma a.2.*[lemmaa2 ] \(i ) since .25mm@xmath688 .25mmis the fourier transform of [email protected] mm , then .25mm@xmath689 is the fourier transform of @xmath690.25 mm , which is [email protected] mm , hence integrable when [email protected] mm .
\(ii ) if @xmath693.25 mm , we simply estimate : @xmath694 if @xmath695.25 mm , we split up @xmath696 using smooth cut off functions @xmath173 and @xmath174 on @xmath175 such that @xmath176.25 mm , @xmath177 on @xmath178 $ ] and @xmath179 on @xmath180.25 mm .
the first integral is estimated as above : @xmath697 after @xmath671 integrations by parts , the second integral becomes @xmath698 hence @xmath699 this concludes the proof of ( ii ) .
the proof of ( iii ) is similar and we omit the details .
* lemma a.3.*[lemmaa3 ] _ assume that @xmath700 where @xmath701.25 mm , .25mm@xmath702 , and @xmath703 .25mmis a symbol of order @xmath704.25 mm . then @xmath705 is a bounded function at the origin .
more precisely , there exists @xmath706 .25mmsuch that , for every @xmath707.25 mm , @xmath708 _ let us split up @xmath709 the first two terms are estimated by integrations by parts .
specifically , @xmath710 with @xmath711 and @xmath712 , while @xmath713 with @xmath714 and @xmath715 .
the last two terms are easy to estimate
. obviously @xmath716.25 mm , .25mmwhile @xmath717 we conclude by summing up these four estimates .
in this appendix we collect some properties of the riesz distributions .
we refer to ( * ? ? ?
* ch.1 , 3 & ch.2 , 2 ) or ( * ? ? ?
* ch.iii , 3.2 ) for more details .
the riesz distribution @xmath718 is defined by @xmath719 when [email protected] mm .
it extends to a holomorphic family .25mm@xmath721 .25mmof tempered distributions on .25mm@xmath34 .25mmwhich satisfy the following properties : * @xmath722 @xmath723.25 mm , * @xmath724 @xmath723.25 mm , * @xmath725 and more generally @xmath726 @xmath727.25 mm , * @xmath728 @xmath729.25 mm .
hence @xmath730 when [email protected] mm .
the riesz distribution .25mm@xmath732 is defined similarly .
their fourier transforms are given by * @xmath733 @xmath734.25 mm , where @xmath735 when @xmath736 and @xmath737 in general ( notice that there are actually no singularities in the last expression ) .
in this appendix we prove the local kernel estimates @xmath738 stated in theorem [ estimatewtildetinfty].i.a under the assumptions [email protected]
mm , @xmath221 .25mmand @xmath739 . by symmetry
, we may assume again that [email protected] mm . by using the first integral representation of the spherical functions in
, we obtain @xmath741 where @xmath742 according to lemma a.2 in appendix a , since .25mm@xmath743 .25mmis a symbol of order .25mm@xmath744
.25mmand @xmath745 where .25mm@xmath671 is the smallest integer @xmath746.25 mm .
hence @xmath747 by using the third integral formula for spherical functions in , we are lead to estimate the expression @xmath749 let us expand @xmath750 as [email protected] mm , and @xmath752 accordingly . since .25mm@xmath753 .25mmis a symbol of order [email protected] mm , its contribution to can be estimated by @xmath755 here we have applied again lemma a.2 and .25mm@xmath671 is the smallest integer @xmath746.25 mm . by using @xmath756 we end up with the estimate @xmath757 notice that the previous computations are valid in dimension [email protected] mm . in dimension .25mm@xmath759 , the last estimate becomes @xmath760 while , in dimension [email protected] mm , is replaced by @xmath761 similarly @xmath762 yields a bounded contribution to .
let us eventually analyze the remaining contribution of @xmath763 which is a classical distribution .
according to the properties of the riesz distributions ( [ riesz ] ) in appendix b , we have indeed @xmath764 and it remains for us to estimate the expression @xmath765 in order to do so , we discuss separately the odd and even dimensional cases .
after .25mm@xmath767 integrations by parts , becomes @xmath768 where .25mm@xmath769 is a linear combination of monomials @xmath770 with @xmath771.25 mm , .25mm@xmath772
.25mmand [email protected] mm . in particular .25mm@xmath774 .
after one more integration by parts , we get @xmath775 where .25mm@xmath776
.25mmand [email protected] mm , hence the last sum is .25mm@xmath778 .25mmand the last integral is [email protected] mm . notice that these terms vanish when [email protected] mm .
thus is [email protected] mm , when .25mm@xmath266 .25mmis odd .
after @xmath767 integrations by parts , becomes this time @xmath782 where .25mm@xmath783 and the other .25mm@xmath769 are as before . since @xmath784 the .25mm@xmath785 .25mmfirst terms in are @xmath786.25 mm . let us turn to the last term @xmath787 which is obtained by taking @xmath788 in .
let us split the integral in ( [ wavekernel5 ] ) as follows : @xmath789 notice that our current assumption @xmath748 .25mmimplies that [email protected] mm .
since @xmath791 the contribution to of the first integral in .25mmcan be estimated by @xmath792 and the contribution to ( [ wavekernel5 ] ) of the last integral in by @xmath793 we handle the remaining integral by performing the change of variables @xmath794 and by integrating by parts @xmath795 all resulting expressions are [email protected] mm , since @xmath797 thus and hence , are [email protected] mm . then , up to a multiplicative constant , @xmath803 let us expand @xmath804 and furthermore @xmath805 the coefficients .25mm@xmath806 .25mmin are constants , while the coefficients .25mm@xmath807 .25mmin are smooth functions on [email protected] mm , which are linear combinations of products @xmath808 with @xmath270.25 mm . consider first @xmath809 since @xmath810 according to the assumption @xmath811 and @xmath812 by taylor s formula , the expression is @xmath813
hence @xmath814 in all cases .
consider next @xmath815 since @xmath816 and @xmath817 the expression is easily seen to be .25mm@xmath818 .25mmas long as [email protected] mm . for the remaining case , where [email protected] mm , let us expand @xmath821 and split @xmath822 in .
on the one hand , the resulting integrals @xmath823 then , up to a multiplicative constant , @xmath831 let us split @xmath832 the following estimate is obtained by resuming the proof of theorem [ estimatewtildetinfty].i.b in the odd dimensional case : @xmath833 since @xmath834 the contribution to of the second integral in is uniformly bounded .
thus we are left with the contribution of the first integral , which is a purely local estimate .
we first prove ( i ) .
recall that @xmath835 hence @xmath836 in both cases . by combining this estimate with @xmath837 and by performing an elementary change of variables ,
we reach our conclusion : @xmath838 we next prove ( ii ) . recall that @xmath839 the remainder s contribution to .25mm@xmath840 .25mmis estimated as above : @xmath841 in order to handle the contribution of .25mm@xmath842 .25mmto [email protected] mm , .25mmlet us perform the change of variables @xmath843.25 mm , .25mmso that @xmath844 and let us expand @xmath845 notice that the expressions @xmath846 .25mmand
.25mm@xmath847 .25mmcan be expanded as follows : @xmath848 @xmath849 @xmath850 using these behaviors and integrating by parts , we can estimate @xmath851 by @xmath852 .
the integrals @xmath853 are estimated similarly . in summary
, we showed that @xmath854 and we conclude by using the behavior of the elementary integral @xmath855 from now on , the discussion of subcase 3.b is similar to subcase 3.a . on the one hand , we deduce from lemma c.1.i that @xmath856 on the other hand , by expanding @xmath857 and @xmath840 .25mmaccording to lemma c.1.ii.25mm , we have @xmath858 where @xmath859 and @xmath860 denote the integrals and , which are uniformly bounded and whose sum is equal to @xmath861 as a conclusion , we obtain again @xmath862 t.h .
koornwinder , _ jacobi functions and analysis on noncompact semisimple lie groups _ , in _ special functions _ ( _ group theoretical aspects and applications _ ) , r.a .
askey & al .
( eds . ) , reidel ( 1984 ) , 185 | we study the dispersive properties of the wave equation associated with the shifted laplace beltrami operator on real hyperbolic spaces and deduce new strichartz estimates for a large family of admissible pairs . as an application
, we obtain local well posedness results for the nonlinear wave equation . |
chemical elements heavier than lithium are thought to be produced exclusively through stellar nucleosynthesis .
the primordial cosmic gas remains chemically pristine until the first supernova ( sn ) explosions expel metals that are produced in the precursor stars .
high - redshift observations of the lyman-@xmath5 forest ( e.g. , songaila 2001 ; pettini et al .
2003 ) , damped lyman-@xmath5 systems ( e.g. , prochaska 2002 ) , and quasars ( e.g. , freudling , corbin , & korista 2003 ) all indicate that heavy elements are distributed in various cosmic environments by @xmath6 .
while the actual transport mechanism and the nature of the primary sources have not been determined , it is often suggested that the origin of these heavy elements may be attributed to the first generation of stars , the so - called population iii stars ( e.g. , ostriker & gnedin 1996 ; madau , ferrara , & rees 2001 ; thacker , scannapieco , & davis 2002 ; qian , sargent , & wasserburg 2002 ) . in the standard cosmological models based on cold dark matter ( cdm ) ,
the first cosmological objects are predicted to form at redshifts greater than 20 ( e.g. , couchman & rees 1986 ; tegmark et al .
1997 ; abel et al . 1998 ; yoshida et al .
thus , the first heavy elements are likely to have been processed at such early epochs . intriguingly , the recent measurement of a large thomson optical depth by the _ wilkinson microwave anisotropy probe ( wmap ) _ satellite ( kogut et al . 2003 ; spergel et al .
2003 ) provides evidence that the universe was reionized very early on , supporting the above notion that the first stars could form at @xmath7 ( e.g. , cen 2003 ; haiman & holder 2003 ; wyithe & loeb 2003a , b ; sokasian et al .
2003a ) .
the relative abundances of heavy elements produced in the first stars are of great importance because observations of the elemental abundance pattern of ultra metal - poor stars ( e.g. , christlieb et al .
2002 ) can place strong constraints on the environments in which these stars were formed and possibly on the progenitor mass ( umeda & nomoto 2003 ; schneider et al .
theoretical modeling of the formation of the first stars ( abel , bryan , & norman 2002 ; bromm , coppi , & larson 2002 ; omukai & palla 2003 ) consistently indicate that they were rather massive , with characteristic mass being possibly several hundred solar masses .
if the first stars are indeed as massive as @xmath8 , they end their lives as energetic sne via the pair - instability mechanism ( e.g. , barkat , rakavy , & sack 1967 ; bond , arnett , & carr 1984 ; fryer , woosley , & heger 2001 ; heger & woosley 2002 ; woosley , heger , & weaver 2002 ) , releasing a total energy of up to @xmath1 ergs . such energetic explosions in the early universe
are violently destructive : they expel the ambient gas out of the gravitational potential well of small - mass dark matter halos , causing an almost complete evacuation ( bromm , yoshida , & hernquist 2003 , hereafter paper i ; wada & venkatesan 2003 ) .
since the massive stars process a substantial fraction of their mass into heavy elements , early sn explosions may provide an efficient mechanism to pollute the surrounding intergalactic medium ( igm ) . models of stellar evolution ( e.g. , heger & woosley 2002 ) predict that , in massive zero - metallicity stars , up to @xmath9% of the total stellar mass is processed into heavy elements in the core and , in the pair - instability case , _ all _ of it is finally ejected during the explosion without leaving any compact remnant behind .
blast waves triggered by the sn explosions will then drive supersonic flows of gas that are enriched with the processed heavy elements .
consequently , because of the short lifetimes of such massive stars , the emergence of the early generation stars should be almost immediately followed by metal enrichment of the igm ; i.e. , prompt metal enrichment could be achieved efficiently by population iii sne in the early universe .
the transport of metals by sn - driven winds has been studied extensively in the context of conventional galaxy formation ( e.g. , larson 1974 ; dekel & silk 1986 ; vader 1986 ; mac low & ferrara 1999 ; aguirre et al .
2001a , b ; madau et al .
2001 ; scannapieco , thacker , & broadhurst 2001 ; mori , ferrara , & madau 2002 ; furlanetto & loeb 2003 ; springel & hernquist 2003a , b ) .
in fact , gas flows induced by sn explosions appear to be the most plausible way of distributing metals , which are observed to exist under a broad range of environments .
interestingly , an array of observations indicate that the metals were produced and dispersed at early stages of structure formation .
spectra of distant quasars consistently show that there are at least some metals in the diffuse igm at @xmath10 ( e.g. , schaye et al . 2000 , 2003 ) .
qian & wasserburg ( 2001 , 2002 ) argue that the observed abundance patterns of metal - poor stars can be explained by the `` prompt - inventory '' of heavy elements from very massive stars .
loewenstein ( 2001 ) points out that a contribution from very massive metal - free stars may account for the observed abundance anomalies of oxygen , silicon , and iron in the intracluster medium ( icm ) .
more recently , freudling et al .
( 2003 ) discovered a strong emission feature from a @xmath11 quasar , providing strong evidence that iron is produced in some stars / galaxies already at @xmath12 , which is expected if black hole growth is regulated by star formation ( e.g. , di matteo et al .
2003a , b ) .
together with the implication of an early reionization derived from the _ wmap _ results , these observations indicate that population iii stars may have been an important contributor of heavy elements in the diffuse igm / icm as well as the source of radiation responsible for the early reionization of the universe .
the association of the igm metal enrichment with the uv - photon production from the first stars was previously discussed by oh et al .
( 2001 ) , and more recently by stiavelli , fall , & panagia ( 2003 ) . in these studies , however , the analysis was not based on a detailed modeling of the high - redshift star formation rate , and the precise epochs of population iii pre - enrichment and reionization therefore remained somewhat uncertain . in the present paper , we revisit the cosmic metal production in the context of early structure formation .
we explore the possibility that the first stars are responsible for reionization of the universe at @xmath13 , while also causing a significant , prompt chemical enrichment of the igm .
we argue that , if the early reionization was the result of very massive population iii stars , the mean metallicity should reach a critical level at which the efficiency of gas cooling is greatly enhanced , thus changing the overall mode of star formation ( e.g. , mackey , bromm , & hernquist 2003 ) .
the hydrodynamic transport of heavy elements induced by sn explosions is studied numerically in paper i , using high - resolution , three - dimensional cosmological simulations .
we show that a large amount of the gas is expelled out of the shallow gravitational potential of the host `` minihalos '' .
the simulations in paper i further demonstrate that , if the explosion kinetic energy is as high as @xmath1 ergs , nearly 90% of the pre - enriched gas in the immediate vicinity of the explosion site eventually escapes out of the host halo , reaching up to a distance of @xmath14 kpc in a few million years .
since multiple sne are expected to be occurring at @xmath13 in distant star - forming regions within a cosmological volume , the total amount of metals produced _ and _ distributed into the igm may well exceed the critical level required to affect the thermal and chemical properties of the igm .
the transition in stellar populations , from population iii to population ii , may then be naturally caused early on . throughout the present paper , except in section [ rsi ] where we consider a variant cdm model
, we work with a flat @xmath15-dominated cold dark matter universe with matter density @xmath16 , cosmological constant @xmath17 and the hubble constant at the present time @xmath18 in units of @xmath19 km s@xmath20mpc@xmath20 .
we set the baryon density to @xmath21 and the initial density fluctuations are normalized to @xmath22
. these parameters are consistent with those employed in the numerical simulations we refer to in the following sections .
in this section we summarize various feedback effects from the first stars , and derive useful order - of - magnitude estimates for quantities needed to study these effects .
the ultimate fate of a metal - free star depends critically on its mass .
theoretical stellar models predict the following classification ( heger & woosley 2002 ; heger et al .
2003 ) : ( 1 ) @xmath23 : these stars explode as core - collapse sne and leave a neutron star behind , ( 2 ) @xmath24 : these explode as faint type ii sne and leave black holes behind after fall - back of most of the envelope ( see umeda & nomoto 2002 , 2003 ) , ( 3 ) @xmath25 : these do not explode as sne and directly collapse into black holes , ( 4 ) @xmath26 : these experience a pulsational instability and eject their outer envelope , again leaving a black hole behind , ( 5 ) @xmath27 : these explode as pair - instability sne , causing complete disruption , and ( 6 ) @xmath28 : these collapse , in the absense of rotation , directly into massive black holes . in summary , only the pair - instability case can substantially contribute to the metal enrichment of the igm , whereas remnant black holes ( or neutron stars ) are left in the other cases , which we shall discuss in 2.5 . while early generation stars may be born with a broad range of masses ( e.g. , bromm , coppi , & larson 1999 ; nakamura & umemura 2001 , 2002 ; omukai & yoshii 2003 ) , very massive stars are dominant in causing various feedback effects to the igm in the early universe such as ionization , heating , and chemical enrichment .
thus , throughout the present paper , we primarily consider very massive ( @xmath29 ) metal - free stars as population iii stars .
we first discuss radiative feedback .
the efficiency of ionizing photon production from massive population iii stars is obtained from detailed calculations ( e.g. , bromm , kudritzki , & loeb 2001b ; schaerer 2002 , 2003 ) . as our fiducial case
, we consider a metal - free star with mass @xmath30 .
such a star emits about @xmath31 ionizing photons per second , and has a main - sequence lifetime of @xmath32 million years .
consider a population of such massive stars formed in a cubic comoving - mpc volume by @xmath33 .
we introduce a quantity @xmath34 where @xmath35 is the cumulative number of ionizing photons emitted from the stars , and @xmath36 is the number of hydrogen atoms in the volume .
this ` number of photons per atom ' @xmath37 serves as an indicator of the global ionization fraction as a function of @xmath33 .
clearly , in order to completely ionize the igm , at least the same number of uv photons as the number of hydrogen atoms must be emitted within the volume .
it follows from these numbers that more than a few hundred very massive stars must be formed in a comoving - mpc volume to cause complete reionization .
massive metal - free stars are not only luminous ionizing sources , but also efficient factories of heavy elements . for stars with masses @xmath29 , typically up to @xmath38% of the mass is processed into heavy elements in the core , consisting mostly of carbon , oxygen , silicon and iron .
we denote this mass fraction with @xmath39 , and set it to be 0.5 as our fiducial value for massive stars in the remainder of the paper .
we further introduce a parameter @xmath40 where @xmath41 is the total number of nuclei in the heavy elements processed in the core , and @xmath42 is the total number of ionizing photons emitted by the star during its lifetime .
namely , with @xmath43 denoting the proton mass , @xmath44 is the total mass of metals produced in the star .
this quantity gives the ratio of the stellar metal to photon production which is , in principle , uniquely determined by the structure and evolution of the respective star . for a metal - free star with mass @xmath45 that is destined to die in a pair - instability supernova ,
the actual value is approximately @xmath46 for @xmath47 .
it is straightforward to compute the value for @xmath48 for less massive stars or for stars with finite metallicity from stellar evolution models .
intriguingly , this parameter @xmath49 for the very massive stars that are destined to explode as pair - instability sne can be expressed in terms of fundamental physical constants as follows .
we estimate the total number of ionizing photons produced during the lifetime of a population iii star with mass @xmath50 as @xmath51 .
here , @xmath52 is the binding energy of an electron in an h atom , with @xmath53 being the bohr radius , @xmath54 the fine - structure constant , and @xmath55 the electron rest mass . assuming the limiting case of a 100% nuclear burning efficiency
, we can write for the total number of nucleons processed into heavy elements : @xmath56 .
we now have , for very massive stars that die as pair - instability sne , @xmath57 where @xmath58 describes the strength of the pion - nucleon interaction .
the production of ionizing photons and metals in a massive star is thus determined by the ratio of the atomic and nuclear binding energies .
the processed metals can be distributed into the igm only if some transport mechanism operates , otherwise the metals would be confined within a small region and the diffuse igm would be largely unaffected . in this section ,
we first describe the hydrodynamics of a single population iii sn explosion in the early universe and then show that the gas and the heavy elements can be transported over a large distance . assuming spherical symmetry of the system , we apply the usual sedov - taylor self - similar analysis for a given explosion energy and density of the ambient medium . we first need to specify the density profile of the surrounding igm in order to fully describe the evolution of the sn blast wave . for a density profile approximated by a simple power - law , @xmath59 , the expansion of a blast wave
is described by the self - similar solution : @xmath60 where @xmath61 is a constant determined by the energy integral , @xmath62 is the explosion energy , @xmath63 the particle number density of the surrounding medium , and @xmath64 the time elapsed since the explosion ( sedov 1959 ; see also ostriker & mckee 1988 ) . in reality , the gas density profile can be more complex because photo - evaporation due to radiation from the central star is very effective in small halos ( @xmath65 k ) .
large pressure gradients in the ionized region drive an outward gas flow _ before _
the explosion is triggered ( see paper i ) .
hence , although initially the gas density is high around the star , it is expected to be substantially reduced by the time when the star explodes , and the density profile will be much shallower except in the outer edge where a supersonic shockwave is propagating . taking the effective photo - evaporation into account , we assume the density of the ambient medium around the explosion site is approximately constant at the mean igm density at @xmath33 , @xmath66 where @xmath67 is the baryon density and @xmath68 is the hubble constant in units of 100 km s@xmath20 mpc@xmath20 .
note that the shock radius in equation ( [ eq : sedov_p ] ) scales only weakly with the density and the explosion energy unless the density profile is very steep .
the three - dimensional simulations in paper i show that the evolution of the shock radius is well described by equation ( [ eq : sedov_p ] ) in a piecewise manner for regions having density profiles with different slopes . for a constant density medium ,
the sedov - taylor self - similar solution reduces to @xmath69 where @xmath70 is the shock radius , @xmath71 the total sn energy in units of 10@xmath72 ergs , @xmath73 the particle number density of the ambient medium in units of 1 @xmath74 , and @xmath75 the elapsed time since the explosion in units of @xmath76 years .
for the explosion energy , heger & woosley ( 2002 ) calculate the energy released by a population iii pair - instability sn to be in the range @xmath77 ergs .
the above solution is valid if the gas remains adiabatic , but breaks down when the shock becomes radiative .
the hot gas in the interior of high redshift ( @xmath12 ) sn remnants ( snrs ) cools partly via inverse compton cooling .
the characteristic cooling time for this process is given by @xmath78 note the strong dependence on redshift .
at @xmath79 , the snr will remain approximately adiabatic ( hence equation [ [ eq : sedov ] ] is valid ) until @xmath80 years after the explosion .
however , this provides only a _ maximum
_ timescale . in the post - shock region ,
the gas density and temperature are so high that hydrogen and helium atomic line cooling is the dominant cooling process .
thus , a thin dense shell could be formed faster than the above timescale .
the exact time when the post - shock region becomes radiative depends on the total explosion energy and the density of the ambient medium in a complicated manner .
the shell forms when the local cooling time @xmath81 at radius @xmath82 becomes shorter than the expansion timescale @xmath83 .
we use the analytic sedov - taylor solution for the temperature @xmath84 and density @xmath85 of the gas to compute the local gas cooling rate @xmath86 , assuming an explosion energy @xmath87 ergs and the gas density @xmath88 @xmath89 ( @xmath90 the mean igm density at @xmath91 ) .
recent numerical simulations by whalen , abel , & norman ( 2003 ) indeed show that photo - evaporation by the central star reduces the final gas density within the host halo to very small values , comparable to the mean igm density . in this case , the shell forms at a time @xmath92 yr . then the shell radius ( the `` stalling - radius '' ) is @xmath93 kpc , and it expands only slowly during the later snow - plow stage as @xmath94 ( see figure 1 in paper i ) .
this is perhaps an optimistic estimate because the gas evacuation efficiency by radiation from the stars depends on the host halo mass and the gas density profile .
for a ten times higher density , @xmath95 @xmath89 , the stalling - radius is calculated to be only 40% smaller , @xmath96 kpc , which is more consistent with the result of our simulation presented in paper i. overall , because of the weak dependence of the self - similar solution ( equation [ [ eq : sedov ] ] ) on the density , the above estimates will not be significantly affected by ( possible ) variations in the gas density around the sn sites .
the momentum - driven slow expansion is still important over a cosmological timescale ( bertschinger 1983 ) , with the radius doubling every twenty folds in time .
thus , if mixing of the original ejecta with the ambient medium is efficient , the expelled metals can travel over a large distance ( @xmath90 a few kpc ) within a hubble time .
three - dimensional numerical simulations of nakasato & shigeyama ( 2000 ) show that the sn ejecta are fairly well mixed within the remnant , with the enriched volume fraction being close to unity .
wada & venkatesan ( 2003 ) show that metal - dispersal is very efficient even in dense , disk - shaped gas clouds .
therefore , the processed metals are likely to be transported as far as the shell radius .
let us calculate the evolution of the volume - averaged mean metallicity of the igm .
we allow the uncertainty associated with the characteristic mass , or initial mass function ( imf ) , of population iii stars by assuming that a fraction , @xmath97 , of the population iii stars have masses in the pair - instability range .
suppose that @xmath98 stars are formed by @xmath33 in a unit comoving volume .
the total mass of metals produced within the volume is then @xmath99 we further normalize this quantity to the mass of the igm within the volume , to obtain the mean metallicity : @xmath100 recall that @xmath49 is approximately a constant for massive metal - free stars that explode as pair - instability sne ( equation [ [ eq : zeta ] ] ) , and @xmath37 traces the igm ionization fraction .
the above equation simply shows that the total quantity of heavy elements ejected into the igm is proportional ( weighted by the fraction @xmath97 ) to the number of uv photons emitted by @xmath33 .
we now consider the simplest case where _ all _ of the population iii stars have a mass of @xmath45 and explode as pair - instability sn , i.e. , @xmath97=1 .
we further assume that @xmath101 marks the epoch of reionization ( see , e.g. , fig .
8 of sokasian et al .
setting @xmath102 , we obtain @xmath103 which coincides with the critical metallicity necessary for the formation of low - mass gas clumps ( bromm et al . 2001a
; see also mackey et al .
note also that very massive stars become progressively unstable for increasing metallicity ( baraffe , heger , & woosley 2001 ) . therefore , for @xmath104 , the epoch when the transition in the mode of star formation occurs should closely match the _ first _ reionization epoch .
note , however , that we have neglected recombination in the above argument .
if population iii star formation occurs rather gradually , taking a considerably longer time than a recombination time , most of the uv photons are simply consumed by recombination without increasing the global ionization fraction of the igm . approximating the igm density with the cosmic mean density
, we estimate the recombination timescale for hydrogen : @xmath105 if the increase in the number of uv photons is sufficiently rapid , compared to @xmath106 , then the ionization fraction of the igm should monotonically increase .
this is indeed the case for the model we describe below .
an important question is _ when and how rapidly _ did the early generation stars emerge ?
we address this issue using the formation rate of population iii stars derived by yoshida et al .
( 2003a ) .
their result is based on a semi - analytic model coupled with the outputs of a large cosmological @xmath107-body simulation . in their model ,
star - forming regions are identified as halos in which the gas has been able to cool via molecular hydrogen cooling , and the usual ` one - star - per - halo ' assumption is made such that only a single star is formed per star - forming gas cloud .
the model also includes the negative feedback owing to photo - dissociation of molecular hydrogen by photons in the lyman - werner band .
hence , the model provides a conservative estimate for the early star - formation rate ( sfr ) .
it is worth noting that the resulting sfr was found to be in good agreement with that obtained from the hydrodynamic simulations of sokasian et al .
( 2003a ) which include radiative transfer as well as various feedback processes .
we revised the empirical fit to the sfr in yoshida et al .
( 2003a ) and found that it is well - described by a particular form proposed by hernquist & springel ( 2003 ) : @xmath108 where @xmath109 and , in this case , @xmath110 , @xmath111 , @xmath112 , @xmath113 , @xmath114 , @xmath115 yr@xmath20 mpc@xmath116 .
figure 1 shows the population iii sfr .
the flattening at @xmath117 reflects the suppression of the gas cooling due to global radiative feedback , showing the self - regulating nature of early star formation ( e.g. , ricotti , gnedin , & shull 2002 ; wise & abel 2003 ; yoshida et al .
we compare the derived sfr with the result of the hydrodynamic simulations of sokasian et al .
( 2003a ) . to this end
, we have averaged over all the nine models considered in sokasian et al .
( 2003a ) which differ in the star - formation efficiency as well as the strengths of the feedback effects .
overall , the agreement between the semi - analytic prediction and the simulation results is quite good . as sokasian et al .
( 2003a ) note , the sfr derived from their simulations may suffer from finite boxsize effects at high redshifts ( @xmath118 ) , which accounts for the small discrepancy between the analytic model and the simulation result seen in figure 1 ( see also yoshida et al .
2003b ; barkana & loeb 2003 ) in the middle panel of figure 1 , we plot the cumulative number of ionizing photons per baryon @xmath37 . we compute the number of ionizing photons assuming a single massive ( @xmath30 ) star emits ionizing photons at a rate of @xmath31 s@xmath20 during its main - sequence lifetime ( bromm et al .
the dashed line in the middle panel indicates @xmath119 , which is achieved by @xmath120 .
we note that the total number of uv photons increases by more than an order of magnitude between @xmath121 and @xmath120 .
the corresponding time interval is @xmath38 myr , whereas the recombination timescale at @xmath91 is @xmath122 myr .
therefore , to lowest order , the photon consumption due to hydrogen recombination can be neglected .
in addition , since the photon escape fraction from minihalos is expected to be of order unity ( whalen et al .
2003 ) , and the clumping of the igm at these early epochs is small ( haiman , abel & madau 1999 ) , @xmath123 a few should suffice to reionize a substantial fraction of the igm ( e.g. , sokasian et al .
it is worth mentioning that , in our model , the time when @xmath119 coincides with the epoch of reionization suggested by the _ wmap _ data , @xmath124 ( assuming an instantaneous reionization , see kogut et al .
2003 ) . at the same time , as shown in the bottom panel in figure 1 , the global mean metallicity reaches the metallicity floor of a few @xmath125 .
we emphasize that this _ global _ mean is merely an approximate estimate because , locally around the sn sites where the formation of second generation stars is likely to take place , the gas metallicity could be much higher than this value . in the previous sections ,
we have exclusively considered very massive stars that explode as pair - instability sne , i.e. , those with masses @xmath126 .
stars with masses outside this range are likely to leave remnant black holes behind ( fryer et al .
in particular , stars even more massive than @xmath127 completely collapse to black holes , which could be the seeds of supermassive black holes ( madau & rees 2001 ; islam , taylor , & silk 2003 ; volonteri , haardt , & madau 2003 ) .
it is straightforward to compute the evolution of the mass density locked up in black holes using our model population iii sfr .
let us adopt the simplest model where all the population iii stars are extremely massive and become black holes .
although this is an implausible situation because then the igm would remain chemically pristine forever and the `` star - formation conundrum '' would arise ( see schneider et al .
2002 ) , it is illustrative to study the evolution of the black hole mass density in this scenario .
we consider three cases with @xmath128 , and the resulting mass density is plotted in figure [ plot2 ] .
note that our model for the population iii star formation is based on the ` one - star - per - halo ' assumption and thus the star formation rate density scales approximately with @xmath129 .
figure [ plot2 ] shows that the mass density at @xmath4 is comparable to the present - day mass density in black holes in the local universe ( e.g. , merritt & ferrarese 2001 ) for high mass values .
note that these population iii blackholes are likely to have sunk to the galactic centers by the present epoch due to dynamical friction ( e.g. , madau & rees 2001 ) .
obviously , this extreme top - heavy mode of population iii star formation should be stopped at some point .
otherwise too much baryonic mass would be converted into black holes locally by @xmath130 .
one may naively expect that population iii star formation could continue down to moderate redshifts , @xmath131 .
we argue , based on the results of our model , that either the completion of reionization or prompt enrichment to a level of @xmath132 by @xmath4 should mark the end of the population iii era .
when a large fraction of the igm is reionized , effective photo - evaporation of minihalos by internal and external sources occurs ( e.g. , shapiro , iliev , & raga 2003 ) , and virtually no gas can be trapped within low mass halos .
furthermore , sn explosions cause nearly complete evacuation of the host halos ( see paper i ) . therefore , regardless of the efficiency of the proposed positive feedback on molecular hydrogen formation ( e.g. , ferrara 1998 ; oh 2001 )
, it should take some time for the heated and expelled gas to cool and _
condense_. subsequent star formation can take place only after the gas falls back into the halos again .
accordingly , there should be an appreciable time gap between the first star formation and the onset of the second - generation star formation .
we conservatively estimate this ` recuperation time ' ( sokasian et al
. 2003a ) to be of the order of the dynamical time scale for the halos .
the dynamical time scale is given by @xmath133 where @xmath134 is the gravitational constant and @xmath135 the total mass density of a halo .
assuming that the mass density of a halo ( including dark matter ) is 200 times the cosmic mean density , we compute the dynamical time for halos collapsing at @xmath136 to be @xmath38 myr . since the cosmic time difference between @xmath137 ( or @xmath138 ) is @xmath139 myr for the standard @xmath15cdm model , star formation is likely to be quenched for at least a duration of @xmath140 a few in redshift .
it might seem that halos with virial temperature greater than @xmath141 k are unaffected by photo - heating due to reionization ( e.g. , oh & haiman 2002 ) . we point out that , in the hierarchical universe , _ all _ the massive halos used to be minihalos at earlier epochs . hence , according to our model of population iii star - formation , the gas within them is likely to have been already expelled either by radiation from the population iii stars or by sn explosions
. the situation would be slightly different in halos that are rapidly growing in mass because such halos may be able to trap the ionized hot gas .
however , sn explosions with energies @xmath1 ergs are still able to cause an outward gas flow even in the large halos . if the feedback from the first massive star on its vicinity occurs by z=15 , as in our model , then the formation of the second - generation stars in the same region ( i.e. , protogalaxy ) will begin at @xmath142 .
further , very massive stars are already pulsationally unstable when the metallicity is @xmath143 , even if they are formed ( baraffe et al .
the minimum metallicity floor is established in all the star - forming regions by @xmath4 if the first stars explode as pair - instability sne , thus causing a ` global ' regulation of the star formation activity lasting until @xmath144 .
note that significant cosmic variance may be induced by large - scale ( @xmath145mpc ) density fluctuations .
the cosmic scatter in the number density of the star - forming clouds should then cause a substantial spread in the epoch of self - termination ( e.g. scannapieco , schneider & ferrara 2003 ) .
an alternative scenario for second - generation star formation is self - propagating star formation in the thin , dense shell formed by early sn explosions ( e.g. , ostriker & cowie 1981 ; wandel 1985 ; mccray & kafatos 1987 ) .
such phenomena are indeed observed in the local universe ( e.g. , dopita et al .
1985 ; efremov & elmegreen 1998 ) and may have taken place in the early universe as well ( mackey et al . 2003 ; salvaterra , ferrara , & schneider 2003 ) . for the particular case we simulated in paper i , we found that no gas clumps are formed in the dense shell which are jeans - unstable .
this may , however , be because of the still insufficient mass resolution of the simulation . in order to properly address the question of the shell fragmentation and the formation of proto - stellar clouds , simulations with an extremely high mass resolution
are needed .
also , more complex configurations , such as collisions between neighboring blast waves originating in more massive systems , may result in the formation of gravitationally bound fragments .
we pursue this possibility further in a forthcoming work ( bromm et al .
, in preparation ) .
in 2 , we have shown that the production of ionizing photons and heavy elements by massive population iii stars is described by fundamental constants in the case where these stars explode as pair - instability supernovae . here , we derive an approximate but concise relationship between the evolution of the mean metallicity and the total thomson optical depth , @xmath146 , as follows . the differential optical depth to thomson scattering corresponding to the redshift interval @xmath147 is @xmath148 where @xmath149 is the thomson scattering cross section , and @xmath150 the mean electron number density at @xmath33 .
the mean electron number density is given by @xmath151 where @xmath152 is the ionization fraction at @xmath33 , and @xmath153 the present - day number density of electrons .
we obtain the contribution from an interval between @xmath154 and @xmath155 ( @xmath156 ) , @xmath157 we have assumed a flat @xmath15cdm model with @xmath158^{1/2}$ ] and the last approximation holds at high redshifts , certainly for @xmath159 .
neglecting recombination , we can approximate , to the lowest order , the ionization fraction as @xmath160 from equation ( [ eq : zmean ] ) we obtain for pair - instability sne , @xmath161 absorbing all the constant factors into a single quantity @xmath162 , including @xmath97 , we express the concise relation as @xmath163 thus , to the lowest order , the evolution of the ionization fraction should be accompanied by the increasing metallicity following equation ( [ eq : tau_z ] ) . integrating equation ( [ eq : tau_z ] ) to @xmath164 , when @xmath165 , we obtain @xmath166 this is the contribution from early population iii stars expected in our model .
metals produced by the early population iii stars should exist ubiquitously in various systems at low redshift .
miralda - escud & rees ( 1997 ) argue that the observed metallicity in the lyman-@xmath5 forest at @xmath168 is consistent with a simple picture in which the metals originate from an early generation of stars .
the average metallicity @xmath169 in the lyman-@xmath5 forest at @xmath168 is still much higher than the mean metallicity achieved at the end of the population iii era , @xmath4 , whereas songaila ( 2001 ) argues that a minimum universal metallicity of about @xmath170 is found in @xmath6 lyman-@xmath5 clouds .
remarkably , this is very close to @xmath171 ( figure 1 ) achieved by the early massive population iii stars .
since the population iii star formation is self - regulating owing to the global radiative feedback and metal enrichment , this value is expected to be approximately the maximum which early population iii stars alone can achieve .
songaila ( 2001 ) estimates the mass density of between @xmath172 and @xmath173 to be @xmath174 .
assuming is the dominant ionization stage with ionization fraction of 0.5 ( songaila 1997 ) , the mass density in carbon is @xmath175 . for a very massive zero - metalicity star ,
the carbon yield is @xmath176 and nearly constant for the relevant core mass ( heger & woosley 2002 ) . from this
we obtain the mass density in carbon ( more precisely , @xmath177c ) for our model with @xmath104 , @xmath178 at @xmath136 .
although smaller than the value that songaila ( 2001 ) obtains for the diffuse igm at @xmath6 , this is not negligible .
indeed , a fraction of the carbon , up to 20% , in the igm may originate from the early generation population iii stars . at @xmath168
, the bulk of the metals found in the lyman-@xmath5 forest may also come from ordinary stellar populations formed in galaxies at lower redshift , @xmath179 ( e.g. , adelberger et al .
interestingly , the metallicity evolution computed by hernquist & springel ( 2003 ) based on their analytic model of star - formation in galaxies predicts that the average metallicity reaches @xmath180 at @xmath181 , in good agreement with the observed value in the lyman-@xmath5 forest .
x - ray observations of galaxy clusters reveal that the icm is enriched to a significant fraction of the solar metallicity ( e.g. , mushotzky et al .
1996 ; white 2000 ) .
in fact , a large fraction ( @xmath182% ) of a typical cluster s endowment with heavy elements exists in the icm rather than in cluster galaxies ( renzini 1997 ) .
clusters retain all the metals produced in them , and the determination of the abundances for several elements in the icm is relatively straightforward , without involving complicated modeling .
thus , the elemental abundances in the icm offer a clean template against which any theoretical models can be tested .
loewenstein ( 2001 ) suggests that pre - enrichment by a generation of massive population iii stars may account for the abundance anomalies of oxygen , silicon and iron , observed in the icm .
the recent analysis of the relative overabundance of si with respect to fe in the icm also indicates the possible contribution from massive population iii stars .
baumgartner et al . ( 2003 ) argue that the observed relative overabundance of si can be explained by population iii progenitors with a 70 @xmath183 he core .
while including the contribution from massive population iii stars is attractive , the required abundance appears very large .
baumgartner et al .
( 2003 ) estimate that the total mass in population iii stars should be at least @xmath184 , in units of the cosmic density parameter , whereas our model predicts only @xmath185 at @xmath136 .
baumgartner et al . ( 2003 ) argue that primordial star formation is enhanced in overdense regions which later become clusters .
however , while this might seem plausible , the strong self - regulation of early star formation primarily tends to shift star formation to earlier epochs in overdense regions , rather than locally enhancing the total amount of formed stars .
we therefore conclude that metals originating from the earliest generation stars can not be responsible for the observed abundance anomalies in the icm .
we have computed the mean metallicity of the igm assuming implicitly that all the metals produced and ejected by sn explosions will remain in the gas phase .
this is not strictly true ; a fraction of the metals will be in the form of dust and hence do not directly affect the gas chemical properties ( see however the effect of promoted molecular hydrogen formation discussed later in this section ) .
schneider , ferrara , & salvaterra ( 2003a ) and nozawa et al . ( 2003 ) calculate the dust depletion factor for pair - instability sne to be up to 30 - 40% .
nozawa et al . ( 2003 ) also conclude that the dust formation rate and the relative abundances of the dust grains of various compounds depend sensitively on the mass of the progenitor and on the mixing efficiency in the ejecta .
another complexity is that the produced dust grains are destroyed by thermal sputtering with ions and electrons , and by grain - grain collisions in post - shock regions ( spitzer 1978 ) .
most important , the reverse shock formed in the early stages of a sn explosion can destroy a significant fraction of the dust grains produced in the ejecta ( mckee 1989 ) .
unfortunately , the efficiency of dust destruction by shocks is not well known and we need to resort to a simple parametrization .
if , for example , the destruction efficiency is very low and nearly all the formed dust grains survive , are dispersed and remain in the igm , the cumulative amount of dust could be significant .
we choose the dust depletion factor to be @xmath187=0.4 ( schneider et al .
2003a ) , and assume a low destruction efficiency , @xmath188=0.1 .
using the global formation rate for population iii stars in equation ( [ eq : sfr_popiii ] ) and assuming @xmath104 , we calculate a conservative upper limit for the mass density of dust in the igm , @xmath189 by @xmath136 .
this is much smaller than , for example , the density of intergalactic dust , @xmath190 at @xmath191 ( aguirre 1999 ; see also pei , fall , & hauser 1995 ; croft et al .
2000 ) , and thus appears to make negligible contribution to the dust in the diffuse igm even if the grains survive until the present epoch .
we point out , however , that even a small quantity of dust in the gas accelerates h@xmath186 production and hence affects the efficiency of gas cooling and star formation when the gas metallicity is still low ( hirashita et al .
todini & ferrara ( 2001 ) argue that h@xmath186 production on dust grains dominates over the usual h@xmath192 channel when the dust - to - gas ratio is greater than 5 percent of the local galactic value @xmath193 .
our model predicts that the mean dust - to - gas ratio is @xmath194 at @xmath136 for @xmath187=0.4 and @xmath188=0.1 , still an order of magnitude smaller than the critical value .
h@xmath186 production on dust grains could still be important if the dust - to - gas ratio were locally high , either because the distribution is inhomogeneous , or because the dust grains remain preferentially within a confined region around the sn site .
it has been suggested that x - rays could act as a source of positive feedback by boosting the free electron fraction in distant regions .
this would in turn promote the formation of h@xmath186 , hence enabling enhanced cooling of primordial gas ( e.g. , haiman , rees , & loeb 1996 ; oh 2001 ; glover & brand 2003 ) .
apart from early quasars ( e.g. , eisenstein & loeb 1995 ; bromm & loeb 2003 ; madau et al .
2003 ) , snrs are the only plausible x - ray sources at very high redshifts .
high redshift snrs cool via inverse - compton cooling , losing a large fraction of their energy to cmb photons ( oh , cooray , & kamionkowski 2003 ) .
some fraction of the thermal energy is also converted into x - rays via thermal bremsstrahlung , and yet another unknown fraction goes into relativistic electrons which could emit x - rays via inverse - compton scattering of cmb photons .
unfortunately , the relative fractions of energy delivered to these processes is highly uncertain , and likely to be time - dependent during the evolution of the snrs . in order to obtain a rough order - of - magnitude estimate
, we calculate the evolution of an early x - ray background using some working assumptions as follows .
we consider only x - rays with energies greater than 1 kev , because the igm is nearly optically thin to these x - rays , and assume further that a fraction @xmath195 of the total sn explosion energy is converted to radiation in the relevant x - ray band , between 1 kev and 5 kev .
we set , somewhat arbitrarily , @xmath196 as our fiducial value .
note that this value is significantly larger than the one assumed by glover & brand ( 2003 ) as a _ maximum _
efficiency case , based on the estimate of terlevich et al .
( 1992 ) for x - ray bright remnants in dense environments , @xmath197 .
hence , the results presented below provide a conservative upper limit . for the emissivity of the snrs , we rather crudely model it to be described by a power law @xmath198 in the narrow ( 1 - 5 kev ) energy range .
by noting that radiative cooling in the snrs occurs mainly in the first few million years for the original sn energy of @xmath1 ergs ( see section 2.3 ) , the time - averaged luminosity in x - rays from a single snr is estimated to be @xmath199 where @xmath200kev .
we compute the evolution of the diffuse x - ray flux by solving the cosmological radiative transfer equation ( e.g. , peebles 1993 ) @xmath201 where @xmath202 is the specific intensity in units of ergs s@xmath20 @xmath203 hz@xmath20 sr@xmath20 , @xmath204 the hubble parameter at redshift @xmath33 , @xmath205 the speed of light , and @xmath206 the proper volume - averaged emissivity .
we use this unconventional unit for the x - ray flux in order to facilitate comparisons with the commonly considered soft - uv background radiation .
we compute the volume emissivity by combining equation ( [ eq : l_nu ] ) with our model sfr ( equation [ 13 ] , see also figure 1 ) , assuming that all the population iii stars die as pair - instability sne ( @xmath104 ) .
this simple model again serves as a maximal case in terms of x - ray emission efficiency .
figure [ fig : xray ] shows the evolution of the x - ray background radiation intensity at 1 kev .
it is clear that the flux reaches only up to @xmath207 ergs s@xmath20 @xmath203 hz@xmath20 sr@xmath20 by @xmath136 .
machacek , bryan , & abel ( 2003 ) performed numerical simulations of pre - galactic object formation to examine the effect of the positive feedback from a diffuse x - ray background in the early universe .
they conclude that the net effect is quite small for x - ray fluxes below @xmath208 ergs s@xmath20 @xmath203 hz@xmath20 sr@xmath20 . to result in a positive feedback that counteracts the negative feedback due to photo - dissociating radiation , an order of magnitude higher flux is necessary , which is two orders of magnitude higher than our estimate for the x - ray flux at @xmath136 .
overall , the effect of the early x - ray background caused by snrs to the primordial gas cooling appears unimportant at @xmath209 .
it has often been suggested that the paucity of metal - poor stars in the solar neighborhood relative to the prediction of theoretical models , the so - called g - dwarf problem , can be resolved by invoking early massive population iii stars ( cayrel 1986 ; see also larson 1998 for a recent review ) .
in fact , the kind of pre - enrichment by early generation stars proposed here may offer an excellent resolution if the metallicity of the gas is sufficiently high initially .
the simplest models require a significant degree of pre - enrichment in the galactic disk ( e.g. , binney & tremaine 1987 ) , @xmath210 , whereas our model predicts far too small a metallicity accomplished by population iii stars alone , @xmath211 .
thus , considering the self - regulating nature of early star formation ( see 3.1 ) , we conclude that massive population iii stars do little to resolve the classical g - dwarf problem .
however , massive population iii stars may account for a modern version of this problem the paucity of extremely metal - poor stars in the galactic halo . as we have shown in previous sections , a prompt enrichment to @xmath211
is achieved by @xmath136 .
in fact , the gas metallicity around the sn sites , where most plausibly the second generation stars will be formed , is higher than this _
mean _ value , and thus , within our model , the second generation stars are expected to have metallicities at or above this critical value .
observationally , with the important exception of the ultra - metal poor star discovered by christlieb et al .
( 2002 ) , the paucity of stars with [ fe / h ] @xmath212 is apparent in our galaxy
. it would be extremely important to derive the abundance pattern of such stars through a systematic search ( e.g. , christlieb 2003 ) .
the rapid metal - enrichment by massive population iii stars has important implications for some variant cdm models .
based on a joint analysis of the first - year _ wmap _ data , the 2df galaxy redshift survey , and the lyman-@xmath5 forest , spergel et al .
( 2003 ) conclude that cosmological models with a running primordial power spectrum are favored over power - law models with @xmath213 ( see also peiris et al .
however , post-_wmap _ studies indicate that the observed high optical depth is not achieved in such models with a strong running , because the formation of pre - galactic size objects is systematically delayed owing to the suppressed power on small scales ( e.g. , somerville et al . 2003 ) .
chiu , fan , & ostriker ( 2003 ) and avelino & liddle ( 2003 ) indeed showed that strong running of the primordial power results in considerable delay of reionization .
yoshida et al . ( 2003b ) carried out cosmological hydrodynamic simulations of early structure formation and concluded that the running spectral index ( rsi ) model is incompatible with the observed high thomson optical depth .
however , yoshida et al . also argue that the resulting thomson optical depth can be made marginally consistent with the _ wmap _ data , _ if _ the ionizing photon production rate is extremely high in galaxies at @xmath214 .
this practically indicates that , in the rsi model , a top - heavy imf for the stars formed in galaxies must be maintained over a long period , @xmath215 .
the problematic nature of this requirement can be appreciated by noting that the total elapsed time to @xmath216 in a @xmath15cdm cosmology is approximately @xmath217 years .
very massive stars , like those considered here , have lifetimes @xmath218 years . thus , if star formation commences at high redshifts , @xmath219 , nearly 500 generations of these high - mass stars could have formed .
as we indicate below , the metals from these stars must be `` hidden '' to an improbably high degree of precision for population iii star formation not to self - terminate well before @xmath216 . here
, we show that , in scenarios such as the running spectral index model , a large quantity of metals is necessarily produced if it is assumed that population iii star formation continues until @xmath216 .
let us consider a simplified model in which the stellar imf is essentially a @xmath220-function at @xmath221 , i.e. , all the stellar baryons are converted into very massive stars .
this is the most efficient way of getting ionizing photons out of a certain mass of baryons converted into stars , because the luminosity per stellar mass is essentially a constant for metal - free stars with mass exceeding @xmath221 . in this case
we may simply assume a high ionizing photon production rate per stellar mass of @xmath222 .
as in yoshida et al .
( 2003b ) , the star - formation rate density for the rsi model is computed using the same functional fit ( equation [ [ eq : sfr_popiii ] ] ) with @xmath223 , @xmath224 , @xmath112 , @xmath113 , @xmath225 , and @xmath226 yr@xmath20 mpc@xmath116 .
figure [ fig : sfr_rsi ] shows the cumulative mass density of stellar baryons .
it directly corresponds to the mass density of black holes if we assume that _ all _ the stars become bhs , i.e. , their masses are greater than @xmath227 . in the plot , the dotted line indicates the present - day mass density in black holes in our local universe estimated by merritt & ferrarese ( 2001 ) .
clearly the mass density increases beyond the _ present - day _ mass density by @xmath228 . on the other hand ,
if all the stars are massive , but within the pair - instability sn range of @xmath229 , up to half of the stellar mass can be processed into heavy elements and expelled .
clearly these models are idealized , but they allow us to explore the possible parameter space . on these assumptions and
using the sfr for the rsi model , we compute the mass - weighted mean metallicity . the bottom panel of figure [ fig : sfr_rsi ] shows the evolution of the mean metallicity @xmath230 .
remarkably , the critical metallicity @xmath143 is achieved already by @xmath231 and it reaches @xmath232 by @xmath173 .
it is now clear that , if we resort to the extremely top - heavy imf in the rsi model to produce enough uv photons to reionize the universe early on , the total amount of heavy elements quickly rises up to a point where the gas chemical properties are expected to change ( bromm et al .
it is conceivable that all the massive stars collapse to black holes so that the igm remains chemically pristine without metal - pollution , but then perhaps too much baryonic mass goes into black holes , which appears to be incompatible with the recent estimate of the local mass density in black holes . in summary
, a running of the primordial power spectrum seems to be incompatible with various observations in one way or another .
a similar conclusion is expected to obtain for other variant cdm models , such as those involving warm dark matter ( e.g. , somerville et al .
2003 ; yoshida et al .
we have revisited the cosmological consequences of an early generation of population iii stars .
we have primarily considered very massive ( @xmath29 ) stars and used a detailed model of the global star formation rate which is derived from semi - analytic modeling and the results of cosmological hydrodynamic simulations that include radiative feedback effects .
we have shown that the first stars alone can produce a sufficient number of photons to reionize a significant volume of the igm by @xmath4 , if the stars are very massive , in agreement with cosmological radiative transfer calculations ( e.g. , sokasian et al .
furthermore , on the assumption that the majority of the stars explode as pair - instability sne , the mean igm metallicity increases to the critical value above which the gas cooling efficiency is greatly enhanced .
this may lead to the formation of ordinary stellar populations including low - mass stars ( e.g. mackey et al .
we also discussed possible implications of the results .
although a broad range of observations indicates that the kind of massive stars we considered in the present paper existed in the early universe , the simple model may not be able to explain the observations in details . clearly , further studies using more sophisticated models are necessary to examine the cosmological consequences of early star formation . in particular
, we will explore the contribution from the second ( and subsequent ) generation stars forming at @xmath233 in future work .
the evidence for an early generation of very massive stars may well be strengthened by future observations .
the ongoing operation of _ wmap _ will yield a more precise value for the total optical depth to reionization . in the longer term ,
post-_wmap _ cmb polarization experiments such as _ planck _ will probe the reionization history ( e.g. , kaplinghat et al .
detection of the second - order polarization anisotropies on arcminute scales can place strong constraint on the details of reionization ( e.g. , liu et al . 2001 ) .
near - infrared observations of afterglows from high - redshift gamma - ray bursts can also be used to probe the reionization history at possibly @xmath159 ( barkana & loeb 2003 ; inoue et al .
2003 ; yoshida & bloom 2003 , in preparation ) .
mapping the morphological evolution of reionization may be possible by observations of redshifted 21 cm emission by the square kilometer array and lofar ( e.g. , tozzi et al .
2000 ; furlanetto , sokasian , & hernquist 2004 ) .
data from these future observations will provide a more complete picture of cosmic reionization and will enable us to distinguish the sources responsible for reionization .
the precise measurement of the near - ir cosmic background radiation will constrain the total amount of light from early generation stars ( santos , bromm , & kamionkowski 2002 ; salvaterra & ferrara 2003 ) . ultimately , direct imaging and spectroscopic observations of high redshift star clusters by the _ james webb space telescope _ will probe
the evolution of stellar populations up to @xmath234 ( e.g. , stiavelli , fall , & panagia 2003 ; tumlinson , shull , & venkatesan 2003 ) .
finally , measurements of the relative abundances of various heavy elements in metal - poor stars should provide valuable information on the chemical history of the universe ( e.g. , burris et al .
2000 ; norris et al .
interestingly , a strong argument against very massive ( @xmath235 ) stars comes from the observed abundance pattern of c - rich , extremely fe - deficient stars ( christlieb et al . 2002 ; umeda & nomoto 2003 ; but see schneider et al .
it remains to be seen whether or not such stars are truly second generation stars and their elemental abundances should precisely reflect the metal - yield from the first sne .
observations of a large number of extremely metal - poor stars will construct better statistics ( norris , ryan , & beers 2001 ) and improve constraints on any models for the early chemical evolution .
understanding the origin of the first heavy elements in the universe and the nature of the sources that are responsible for cosmic reionization will require the concerted use of data from these broad classes of observations .
we thank evan scannapieco for helpful comments , and aaron sokasian for providing us with the sfr data .
ny acknowledges support from the japan society of promotion of science special research fellowship ( grant 02674 ) . this work was supported in part by nsf grants aci 96 - 19019 , ast 98 - 02568 , ast 99 - 00877 , and ast 00 - 71019 and nasa atp grant nag5 - 12140 .
mckee , c. , 1989 , in interstellar dust : proceedings of the 135th symposium of the international astronomical union , eds .
l. j. allamandola and a. g. g. m. tielens .
135 , kluwer academic publishers ( dordrecht ) , 431 | the birth and death of the first generation of stars have important implications for the thermal state and chemical properties of the intergalactic medium ( igm ) in the early universe .
sometime after recombination , the neutral , chemically pristine gas was reionized by ultraviolet photons emitted from the first stars , but also enriched with heavy elements when these stars ended their lives as energetic supernovae . using the results from previous high - resolution cosmological simulations of early structure formation that include radiative transfer ,
we show that a significant volume fraction of the igm can be metal - polluted , as well as ionized , by massive population iii stars formed in small - mass ( @xmath0 ) halos early on .
if most of the early generation stars die as pair - instability supernovae with energies up to @xmath1 ergs , the volume - averaged mean metallicity will quickly reach @xmath2 by a redshift of @xmath3 , possibly causing a prompt transition to the formation of a stellar population that is dominated by low - mass stars . in this scenario ,
the early chemical enrichment history should closely trace the reionization history of the igm , and the end of the population iii era is marked by the completion of reionization and pre - enrichment by @xmath4 .
we conclude that , while the pre - enrichment may partially account for the `` metallicity - floor '' in high - redshift lyman-@xmath5 clouds , it does not significantly affect the elemental abundance in the intracluster medium . |
the dynamics of particles dispersed in a host solvent , how they react to and affect the fluid motion , is a relevant problem for science as well as engineering fields@xcite .
it is necessary to account for the macroscopic properties of suspensions ( such as the viscosity , elastic modulus , and thermal conductivity ) , as well as the mechanics of protein unfolding@xcite , the kinetics of bio - molecular reactions@xcite , and the tumbling motion of bacteria@xcite .
as the particles move , they generate long - range disturbances in the fluid , which are transmitted to all other particles . properly accounting for these hydrodynamic interactions
has proven to be a very complicated task due to their non - linear many - body nature .
several numerical methods have been developed to explicitly include the effect of hydrodynamic interactions in a suspension of particles@xcite .
however , their applicability to non - newtonian host solvents or solvents with internal degrees of freedom is not straightforward , and in some cases not possible .
we have proposed an alternative direct numerical simulation ( dns ) method , which we refer to as the smooth profile ( sp ) method , that simultaneously solves for the host fluid and the particles .
the coupling between the two motions is achieved through a smooth profile for the particle interfaces .
this method is similar in spirit to the fluid particle dynamics method@xcite , in which the particles are modeled as a highly viscous fluid .
the main benefit of our model is the ability to use a fixed cartesian grid to solve the fluid equations of motion .
the sp method has been successfully used to study the diffusion , sedimentation , and rheology of colloidal dispersions in incompressible fluids@xcite .
recently it has been extended to include self - propelled swimmers@xcite and for compressible host solvent@xcite .
so far , however , only spherical particles were considered . in this paper
we extend the sp method to be applicable to arbitrary rigid bodies .
we show the validity of the method by computing the mobility / friction tensors for a large variety of geometric shapes .
the results are compared to numerical solutions of the stokes equation , which are essentially exact , as well as experimental data .
the agreement with our results is excellent in all cases considered here .
future papers in this series will deal with the dynamical properties of rigid body dispersions in detail , with this work we aim to introduce the basics of the model and show its validity .
we solve the dynamics of a rigid body in an incompressible newtonian host fluid using the sp method .
the basic idea behind this method is to replace the sharp boundaries between the particles and the fluid with a smooth interface .
this allows us to define all field variables over the entire computational domain , and results in an efficient method to accurately resolve the many - body hydrodynamic interactions .
the motion of the host fluid is governed by the incompressible navier - stokes equation@xcite @xmath0 where @xmath1 is the fluid velocity , @xmath2 the density , and @xmath3 the newtonian stress tensor @xmath4\label{e : dstress}\end{aligned}\ ] ] with @xmath5 and @xmath6 the pressure and viscosity of the fluid , respectively .
the motion of the particle is given by the newton - euler equations@xcite , @xmath7 where @xmath8 , @xmath9 , @xmath10 , and @xmath11 , are the center of mass ( com ) position , velocity , angular velocity , and orientation matrix , respectively , of the @xmath12-th rigid body ( @xmath13 ) .
the total force and torque experienced the particles is denoted as @xmath14 and @xmath15 , respectively , with @xmath16 the moment of inertia tensor and @xmath17 the skew - symmetric angular velocity matrix @xmath18 the force ( and torque ) on each of the particles @xmath19 is comprised of hydrodynamic contributions @xmath20 , particle - particle interactions ( including a core potential to prevent overlap ) @xmath21 , and a possible external field contribution @xmath22 .
for the present study , we neglect thermal fluctuations , but they are easily included within the sp formalism@xcite .
the coupling between fluid and particles is obtained by defining a total velocity field @xmath23 , with respect to the fluid @xmath1 and particle @xmath24 velocity fields , as @xmath25 \label{e : up}\end{aligned}\ ] ] where @xmath26 is a suitably defined sp function ( @xmath27 ) that interpolates between fluid and particle domains ( as described below ) , and @xmath28 . the modified navier - stokes equation which governs the evolution the total fluid velocity field ( host fluid @xmath29 particle )
is given by@xcite @xmath30 with @xmath31 and @xmath32 .
the stress tensor is defined as in eq . , but in terms of the total fluid velocity @xmath23 .
the scheme used to solve the equations of motion is the same fractional - step algorithm introduced in ref@xcite .
, with minor modifications needed to account for the non - spherical geometry of the particles .
let @xmath33 be the velocity field at time @xmath34 ( @xmath35 the time interval ) .
we first solve for the advection and hydrodynamic viscous stress terms , and propagate the particle positions ( orientations ) using the current particle velocities @xmath36 \label{e : ustar}\\ { \bm{r}}_i^{n+1 } & = { \bm{r}}_i^n + \int_{t_n}^{t_n+h}{\text{d}s\,}{\bm{v}}_i \label{e : rn } \\ { { \bm{\mathsf{q}}}}_i^{n+1 } & = { { \bm{\mathsf{q}}}}_i^{n } + \int_{t_n}^{t_n+h}{\text{d}s\,}\text{skew}\left({\bm{\omega}}_i\right)\cdot{{\bm{\mathsf{q}}}}_i\label{e : qn}\end{aligned}\ ] ] given the dependence of the profile function on the particle position and orientation , we must also update the particle velocity field to @xmath37\end{aligned}\ ] ] next , we compute the hydrodynamic force and torque exerted by the fluid on the particles , by assuming momentum conservation .
the time integrated hydrodynamic force and torque over a period @xmath35 are equal to the momentum exchange over the particle domain @xmath38 & = \int{\text{d}{\bm{x}}\,}\rho\phi_i^{n+1}\left({\bm{u}}^ * - { \bm{u}}_p^{*}\right )
\\ \left[\int_{t_n}^{t_n+h}{\text{d}s\,}{\bm{n}}_i^{\text{h}}\right ] & = \int{\text{d}{\bm{x}}\,}\left[{\bm{r}}_i^{n+1}\times \rho\phi_i^{n+1}\left({\bm{u}}^ * - { \bm{u}}_p^*\right)\right]\end{aligned}\ ] ] from this , and any other forces on the rigid bodies , we update the velocities of the particles as @xmath39 \\ { \bm{\omega}}_i^{n+1 } & = { \bm{\omega}}_i^{n } + { { \bm{\mathsf{i}}}}_i^{-1}\int_{t_n}^{t_n+h}\cdot\left[{\bm{n}}_i^{\text{h } } + { \bm{n}}_i^{\text{c } } + { \bm{n}}_i^{\text{ext}}\right]\end{aligned}\ ] ] finally , the particle rigidity is imposed on the total fluid velocity through the body force in the navier - stokes equation @xmath40 \\
\left[\int_{t_n}^{t_n+h}{\text{d}s\,}\phi{\bm{f}}_p\right ] & = \phi^{n+1}\left({\bm{u}}_p^{n+1 } - { \bm{u}}^{*}\right ) - \frac{h}{\rho}\nabla p_p\end{aligned}\ ] ] with the pressure due to the rigidity constraint obtained from the incompressibility condition @xmath41 . for computational simplicity
, we consider each particle ( @xmath12 ) as being composed of a rigid collection of @xmath42 spherical beads ( see fig.[f : beads ] ) , with position , velocities , and angular velocities given by @xmath43 , @xmath44 , and @xmath45 .
we use upper and lowercase variables to differentiate between rigid body particles and the spherical beads used to construct them , as well as the shorthand @xmath46 to refer to the @xmath42 beads belonging to the rigid body @xmath12 . the rigidity constraint on the bead velocities
is given by @xmath47 where @xmath48 is the distance vector from the com of the rigid body ( @xmath8 ) to the bead s com ( @xmath43 ) .
the rigidity constraint on the position of the beads requires that the relative distances between any two of them remain constant .
thus , the @xmath49 vectors , expressed within the reference frame of the particle @xmath50 , are constants of motion @xmath51 where @xmath52 is the matrix transpose of @xmath53 .
the individual positions of the beads can be directly obtained from the position and orientation of the rigid body to which they belongs through @xmath54 rigid body representation as arbitrary collection of spherical beads.,scaledwidth=35.0% ] the beads should only be considered as a computational bookkeeping device , used to map the rigid particles onto the computational grid used to solve the fluid equations of motion .
we are free to chose any representation of the rigid body .
the advantage of this spherical - bead representation is the ease with which the smooth profile function @xmath55 of an arbitrary rigid body can be defined .
we start with the profile function for spherical particles introduced in ref .
@xcite @xmath56 } { h\left[\left(a+\xi/2\right ) - r_i\right ] + h\left[r_i - \left(a - \xi/2\right)\right]}\\ h(x ) & = \begin{cases } \exp{\left(-\delta^2 / x^2\right ) } & x \ge 0 \\ 0 & x < 0 \end{cases}\end{aligned}\ ] ] where @xmath57 is the distance vector from the sphere center to the field point of interest , @xmath58 is the radius of the spheres , and @xmath59 is the thickness of the fluid - particle interface .
we then define the smooth profile function of the rigid body as @xmath60 the normalization factor in eq .
is required to avoid double - counting in the case of overlap between beads belonging to the same rigid particle ( beads belonging to different particles are prevented from overlapping by the core potential in @xmath61 ) .
we note that this representation of particles as rigid assemblies of spheres does not impose any constraints on the particle geometries , because the constituent beads are free to overlap with each other .
if the dispersion under consideration is such that the inertial forces are negligible compared to the viscous forces , i.e. when the particle reynolds number is vanishingly small @xmath62 ( @xmath63 and @xmath64 being the characteristic velocity and length scales ) , the incompressible navier - stokes equation ( eq . ) reduces to the stokes equation@xcite @xmath65 with @xmath66 any external forces on the fluid . due to the linear nature of this equation , the force ( torque ) exerted by the fluid on the particles is also linear in the their velocities@xcite @xmath67 where @xmath68 and @xmath69 are @xmath70 dimensional force and velocity vectors , and @xmath71 and @xmath72 are the friction and mobility matrices , respectively @xmath73 { { \bm{\mathsf{\zeta}}}}^{rt}_{11 } & \cdots & { { \bm{\mathsf{\zeta}}}}^{rt}_{1n } & { { \bm{\mathsf{\zeta}}}}^{rr}_{11 } & \cdots & { { \bm{\mathsf{\zeta}}}}^{rr}_{1n } \\
\vdots & \ddots & \vdots & \vdots & \ddots & \vdots \\ { { \bm{\mathsf{\zeta}}}}^{rt}_{n1 } & \cdots & { { \bm{\mathsf{\zeta}}}}^{rt}_{nn } & { { \bm{\mathsf{\zeta}}}}^{rr}_{1n } & \cdots & { { \bm{\mathsf{\zeta}}}}^{rr}_{nn } \end{pmatrix}\end{aligned}\ ] ] @xmath74 { { \bm{\mathsf{\mu}}}}^{rt}_{11 } & \cdots & { { \bm{\mathsf{\mu}}}}^{rt}_{1n } & { { \bm{\mathsf{\mu}}}}^{rr}_{11 } & \cdots & { { \bm{\mathsf{\mu}}}}^{rr}_{1n } \\
\vdots & \ddots & \vdots & \vdots & \ddots & \vdots \\ { { \bm{\mathsf{\mu}}}}^{rt}_{n1 } & \cdots & { { \bm{\mathsf{\mu}}}}^{rt}_{nn } & { { \bm{\mathsf{\mu}}}}^{rr}_{1n } & \cdots & { { \bm{\mathsf{\mu}}}}^{rr}_{nn } \end{pmatrix}\end{aligned}\ ] ] where the off - diagonal matrices are related through the lorentz reciprocal relations by@xcite @xmath75 the symmetric block matrices @xmath71 ( @xmath76 ) are composed of @xmath77 friction ( mobility ) matrices @xmath78 , @xmath79 , @xmath80 , and @xmath81 which couple the translational and rotational motion of particle @xmath82 with that of particle @xmath83 .
thus , the whole problem of solving the dynamical equations of motion for a suspension of spheres in the stokes regime reduces to calculating the mobility or friction matrix . for a single spherical particle , translating ( rotating ) in an unbounded fluid under stick - boundary conditions , the translational @xmath84 and rotational @xmath85 friction coefficients are obtained from an exact solution of the stokes equation as@xcite @xmath86 exact solutions to the friction or resistance problem of two or three spherical particles are known@xcite , but for arbitrary many - particle systems , the complex non - linear nature of the hydrodynamic interactions makes it impossible to find a general solution .
however , several methods have been developed to obtain accurate estimates for the mobility and friction matrices .
two of the most popular approaches are the method of reflections and the method of induced forces@xcite .
the former relies on a power series expansion of the flow field , in terms of the inverse particle distances ( @xmath87 ) , while the latter uses a multipole expansion of the force densities induced at the particle surface , with the truncation scheme determined by the angular dependence of the flow field ( @xmath88 ) . as an example , the popular rotne - prager ( rpy ) approximation to the mobility tensor can be obtained using the method of reflections , by truncating the hydrodynamic interactions to third order , which corresponds to a pair - wise representation @xmath89 & i\ne j \end{cases}\end{aligned}\ ] ] we will compare our dns results with those obtained using the method of induced forces , using the freely available ` hydrolib ` library@xcite .
calculations of the friction coefficients for a sedimenting array of spherical clusters , using the induced force method truncated to third order ( @xmath90 ) , gives an error of less than @xmath91 with respect to the experimental results@xcite .
we therefore consider these values as the exact solution to the stokes equation ( se ) .
finally , we note that neither the method of reflections nor the induced force method is able to directly take into account lubrication forces , caused by the relative motion of particles at short distances , as they require the high - order terms to be included in both expansions .
following durlofsky et al.@xcite , these contributions are usually added to the friction matrix by assuming a pairwise superposition approximation . in this work
we will consider only the collective motion of a rigid agglomerate of spheres , so that lubrication effects need not be considered .
in what follows we report the friction coefficients for a variety of non - spherical particles under steady translation ( rotation ) through a fluid .
the numerical simulations are performed in three dimensions under periodic boundary conditions . the modified navier - stokes equation ( eq . )
is discretized with a dealiased fourier spectral scheme in space and an euler scheme in time .
the motion of the particles is integrated using a second order adams - bashforth scheme .
the lattice spacing @xmath92 is taken as unit of length , the unit of time is given by @xmath93 , with @xmath94 and @xmath95 the density and viscosity of the fluid .
the integration time step is @xmath96 .
we only consider neutrally bouyant particles , @xmath97 , so gravity effects are not considered furthermore , we are only interested in single particle motion , so the particle - particle interactions and external field contributions can be ignored @xmath98 . the rigid particles are constructed as rigid agglomerates of non - overlapping spherical beads of equal radius @xmath58 .
we use a bead radius of @xmath99 and a system size of @xmath100 ( depending on the particle geometry ) ; the interfacial thickness is @xmath101 in all cases .
the particle velocity @xmath102 is fixed throughout the simulation and the steady - state forces @xmath103 , which are purely hydrodynamic in nature , are measured in order to obtain the friction coefficients @xmath104 .
all results are given in terms of the kinetic form factors @xmath105 , defined as @xmath106 such that @xmath107 ( @xmath108 ) expresses the force ( torque ) on the agglomerates relative the force ( torque ) experienced by a spherical particle of equal volume ( radius @xmath109 ) moving ( rotating ) at the same velocity ( angular velocity ) and under the same boundary conditions .
we use the general term , friction tensor or matrix , to refer to both @xmath71 and @xmath110 .
0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig:",scaledwidth=85.0% ] 0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig : " ] 0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig : " ] 0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig : " ] 0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig : " ] 0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig : " ] 0.15 ( color online ) closed - packed ( hcp and fcc ) arrays of spherical particles .
colors are assigned to the individual layers as a visual guide.,title="fig : " ] we begin by measuring the friction coefficients of the spherical agglomerates studied experimentally in ref @xcite . the agglomerates where constructed to obtain semi - spherical geometries by gluing together from two to @xmath111 spherical beads of equal radius within a closest - packing arrangement ( using either hcp or fcc lattices ) .
the spherical agglomerates for @xmath112 are shown in fig .
[ f : spherical ] .
the system parameters are @xmath113 , @xmath114 , and @xmath115 ( @xmath116 ) , which gives a reynolds number of @xmath117 . given the symmetry of the particles , the friction matrix is diagonal , with only two distinct coefficients @xmath118 ( color online ) friction coefficients @xmath119 for several closed - packed arrays of spherical particles as a function of size ( number of spheres ) .
results obtained from dns calculations ( circle ) are compared to the exact solution to the stokes equation ( square).,scaledwidth=48.0% ] .[t : spherical ] friction coefficients for motion parallel and perpendicular to the vertical @xmath120-axis for all the closed - packed arrays , as given by our dns method and the exact solution to the se . [ cols="<,^,^,^,^,^,^",options="header " , ] precise experimental measurements are available for these systems , but they should not be compared directly to our simulation results due to the mismatch in boundary conditions , particularly for the larger agglomerates .
the friction coefficients for movement along the vertical @xmath120-axis are given in fig [ f : spherical_k ] , where they are compared to the exact ( se ) results .
the complete set of values for the two independent friction coefficients are given in table [ t : spherical ] .
for the larger systems , @xmath121 , the ` hydrolib ` library does not converge if periodic boundary conditions are used , so no reference data is given .
our results show excellent agreement with the available se values , differentiating between nearly identical agglomerates which differ in volume by only a few percent . in all cases , the difference between our results and the reference values are less than @xmath122 , which is the comparable with the error estimates of the actual experiments .
we now consider the friction coefficients for a series of non - spherical regular - shaped agglomerates .
the simulation protocol is exactly the same as for the spherical agglomerates considered above . in total
, we study six different families of configurations , shown schematically in fig .
[ f : nonspherical ] , v - shaped , w - shaped , h - shaped , hexagonal and rectangular arrays .
the v- and w - shaped configurations vary in the number of particles ( @xmath123 ) , as well as the branching angle @xmath124 . for the h - shaped and hexagonal configurations only the branching angle
is varied , the number of particles is fixed to six . for the rectangular @xmath125 arrays , the maximum number of particles in any dimension is four . in total
, we have considered @xmath126 different geometric configurations .
details on the construction of the agglomerates , as well as experimental data for the kinematic form factors , can be found in ref .
@xcite . 0.2 ( color online ) non - spherical agglomerates for various regular shaped geometries.,title="fig : " ] 0.2 ( color online ) non - spherical agglomerates for various regular shaped geometries.,title="fig : " ] 0.2 ( color online )
non - spherical agglomerates for various regular shaped geometries.,title="fig : " ] 0.2 ( color online ) non - spherical agglomerates for various regular shaped geometries.,title="fig : " ] 0.2 ( color online ) non - spherical agglomerates for various regular shaped geometries.,title="fig : " ] 0.2 ( color online ) non - spherical agglomerates for various regular shaped geometries.,title="fig : " ] the total friction matrix @xmath110 is block diagonal for all the geometrical configurations considered here , except for the case of v- and w - shaped particles , for which a slight coupling between translation and rotation can be observed : @xmath127 and @xmath128 .
for the moment , however , we consider only translational motion . due to the symmetry of the particles ,
the friction matrix @xmath107 is again diagonal @xmath129 we have computed the friction coefficients for motion parallel to the vertical @xmath120 axis for all the systems , the full friction matrix is only measured for the h - shaped and hexagonal agglomerates .
our results are summarized in fig .
[ f : nonspherical_k ] , along with the experimental data , and the exact solutions for both a periodic and an infinite system .
although our dns results should only be compared with the se solutions under equivalent boundary conditions , the periodicity effects for the systems considered here are small , being of the same order of magnitude as the errors in the experiments .
the relative error of the dns results ( compared to experiments ) is less than @xmath130 for all configurations .
a comparison of our results with the exact se values under periodic boundary conditions show almost perfect agreement , considerably better than that of experiments with the exact se values for an infinite system .
0.42 ( color online ) friction coefficients for various non - spherical regular - shaped agglomerates .
( a ) all the friction coefficients computed with the dns method as a function of the exact se values .
( b ) the friction coefficients for the hexagonal shaped agglomerates as a function of their vertical dimension .
( c ) vertical friction coefficients for rectangular @xmath125 arrays as a function of height @xmath131 .
values obtained from the exact solution to the se , for both periodic and unbounded @xmath132 systems , as well as experimental values , are also shown in ( b ) and ( c).,title="fig : " ] 0.42 ( color online ) friction coefficients for various non - spherical regular - shaped agglomerates .
( a ) all the friction coefficients computed with the dns method as a function of the exact se values .
( b ) the friction coefficients for the hexagonal shaped agglomerates as a function of their vertical dimension .
( c ) vertical friction coefficients for rectangular @xmath125 arrays as a function of height @xmath131 .
values obtained from the exact solution to the se , for both periodic and unbounded @xmath132 systems , as well as experimental values , are also shown in ( b ) and ( c).,title="fig : " ] 0.42 ( color online ) friction coefficients for various non - spherical regular - shaped agglomerates . ( a )
all the friction coefficients computed with the dns method as a function of the exact se values .
( b ) the friction coefficients for the hexagonal shaped agglomerates as a function of their vertical dimension .
( c ) vertical friction coefficients for rectangular @xmath125 arrays as a function of height @xmath131 .
values obtained from the exact solution to the se , for both periodic and unbounded @xmath132 systems , as well as experimental values , are also shown in ( b ) and ( c).,title="fig : " ] the friction coefficients @xmath133 of all the configurations are plotted in fig .
[ f : nonspherical_k_all ] with respect to the exact se value for a system with the same periodic boundary conditions @xmath134 .
the results clearly show the accuracy of our method , as @xmath135 within @xmath136 .
detailed results for the hexagonal particles are given in fig .
[ f : nonspherical_k_hex ] , where the three form factors @xmath137 are plotted as a function of the vertical dimensions of the agglomerate ( @xmath138 ) .
the dns results show almost perfect agreement with the exact se results .
the difference among the frictions coefficients and their dependence on the branching angle is very accurately reproduced .
finally , the form factors for the rectangular @xmath125 arrays are plotted in fig .
[ f : nonspherical_k_rect ] , as a function of increasing vertical height @xmath131 .
as expected , the agreement with the exact results is very good , and we are able to accurately distinguish between particles with the same cross - sectional area @xmath139 .
( color online ) helical structures of varying pitch @xmath140 ( @xmath141 is the particle diameter ) , with constant length ( @xmath142 beads ) and number of turns ( @xmath143).,scaledwidth=36.0% ] up to now , we have considered symmetric particles for which the translational and rotational motion are only weakly coupled , if at all . here
we will analyze chiral structures which exhibit a very strong coupling between the two motions .
we consider left - handed helices composed of a fixed number of beads ( @xmath142 ) and turns ( @xmath143 ) , which vary only in their degree of pitch @xmath144 ( surface to surface distance between turns ) .
the pitch will then vary in the range @xmath145 , with @xmath146 corresponding to a packed configuration ( hollow cylinder ) , and @xmath147 to a linear chain ( see fig .
[ f : helix ] ) .
the simulation protocol is slightly modified with respect to the previous systems , we now consider beads of radius @xmath148 and a simulation box size of @xmath149 . to compute the complete friction matrix , we now require six different simulations for each helical geometry : three at fixed velocity @xmath150 and three at fixed angular velocity @xmath151 ( @xmath152 ) .
the friction matrices obtained from the dns simulations , together with the exact se values , for a helix with pitch @xmath153 ( @xmath154 the diameter of the beads ) are given in eqs .
@xmath155 @xmath156 @xmath157 the complex nature of the fluid flow generated by the motion of the body is clearly evident in the form of the friction matrices . the translational ( rotational ) friction matrix @xmath107 ( @xmath108 )
is no longer diagonal , eqs . and , although it remains symmetric , which means that the hydrodynamic force ( torque ) will not be parallel to the direction of motion ( axis of rotation ) .
we note that although the off - diagonal components can be up two three orders of magnitude smaller than the diagonal components , the dns method is able to accurately measure all contributions .
although this accuracy is slightly reduced when considering the coupling between translation and rotation @xmath158 ( the small off - diagonal components can show large relative errors @xmath159 ) , the dominant components are well reproduced . to establish a clear estimate of the error
, we use the frobenius norm @xmath160 as a measure of the difference between the two matrices @xmath161 the overall error of the dns method , computed as the relative distance between the dns and se friction matrices is @xmath162 .
@xmath163 @xmath164 friction coefficients for a helix of fixed length and number of turns , as a function of pitch @xmath165 ( @xmath141 is the bead radius ) .
( top ) rotation - rotation and ( bottom ) rotation - translation friction tensor .
the results obtained from dns simulations are compared to the exact se values , as well as those obtained using an rpy approximation.,title="fig:",scaledwidth=48.0% ] + @xmath164 friction coefficients for a helix of fixed length and number of turns , as a function of pitch @xmath165 ( @xmath141 is the bead radius ) .
( top ) rotation - rotation and ( bottom ) rotation - translation friction tensor .
the results obtained from dns simulations are compared to the exact se values , as well as those obtained using an rpy approximation.,title="fig:",scaledwidth=48.0% ] + finally , to see in what way the structure of the helix will affect its motion , we consider the @xmath164 components of the @xmath166 and @xmath108 friction matrices , as a function of the pitch distance @xmath144 .
as the pitch is increased and the helix starts to stretch , fluid flow between the turns of the helix will start to increase , while the cross sectional area of the particle ( @xmath167 plane ) is reduced .
the latter will increase the torque felt by the helix ( as the particle must now drag the fluid along ) , while the former will tend to reduce it ( by reducing the moment of inertia ) .
these effects give rise to the behavior seen in fig [ f : helix_pitch ] , where the @xmath108 coefficient shows a maximum at an intermediate pitch value @xmath168 .
similar behavior is observed for the @xmath166 coefficient , although the maximum is obtained at a different pitch value @xmath169 , and the coupling between rotation and translation disappears for @xmath170 and @xmath171 , as expected . for comparison purposes ,
we have also plotted the results obtained using the ewald - summed rpy tensor@xcite in fig [ f : helix_pitch ] .
the agreement is surprisingly good for the @xmath158 tensor , but a considerable discrepancy appears in the coefficients of the @xmath108 tensor , as this rpy formulation does not include hydrodynamic effects arising from the particle rotation .
we have extended the sp method to apply to arbitrary rigid bodies . for the moment ,
we have considered only particles constructed as a rigid agglomerate of ( possibly overlapping ) spherical beads , but alternative formulations are straightforward .
we have verified the accuracy of our method by performing low reynolds number dns simulations to compute the single particle friction coefficients for a large variety of rigid bodies .
our results were compared with the exact solutions to the stokes equation , showing excellent agreement in all cases . while there are several methods capable of performing these type of calculations , they impose a number of restrictions which can severely limit the type of systems to which they can be applied .
our method can capture the many - body hydrodynamic effects very accurately , and it is not restricted to zero reynolds number flow or newtonian host solvents . in future papers
we will consider the lubrication effects of non - spherical particles ( chains , rods , disks , tori , etc ) , as well as the dynamics at high reynolds number and in the presence of a background flow fields .
the authors would like to express their gratitude to the japan society for the promotion of science for financial support ( grants - in - aid for scientific research kakenhi no .
23244087 ) and mr .
takuya kobiki for his help with the simulations on helical particles .
34ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://www.amazon.com/colloidal-dispersions-cambridge-monographs-mechanics/dp/0521426006%3fsubscriptionid%3d1v7vtj4ha4mft9xbj1r2%26tag%3dmekentosjcom-20%26linkcode%3dxm2%26camp%3d2025%26creative%3d165953%26creativeasin%3d0521426006[__ ] , ed .
( , , ) link:\doibase 10.1063/1.1673799 [ * * , ( ) ] link:\doibase 10.1063/1.3665415 [ * * , ( ) ] http://prl.aps.org/abstract/prl/v109/i24/e248109 [ * * , ( ) ] http://gateway.webofknowledge.com/gateway/gateway.cgi?gwversion=2&srcauth=mekentosj&srcapp=papers&destlinktype=fullrecord&destapp=wos&keyut=a1988n090500008 [ * * , ( ) ] link:\doibase 10.1063/1.466366 [ * * , ( ) ] link:\doibase 10.1063/1.478857 [ * * , ( ) ] link:\doibase 10.1006/jcph.2000.6592 [ * * , ( ) ] link:\doibase 10.1006/jcph.2000.6542 [ * * , ( ) ] http://link.springer.com/article/10.1023/a:1010414013942 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/16/38/009 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.85.1338 [ * * , ( ) ] link:\doibase 10.1103/physreve.79.031401 [ * * , ( ) ] link:\doibase 10.1103/physreve.81.041807 [ * * , ( ) ] link:\doibase 10.1103/physreve.80.061402 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physreve.84.051404 [ * * , ( ) ] link:\doibase 10.1103/physreve.87.022310 [ * * , ( ) ] link:\doibase 10.1039/c3sm00140 g [ * * , ( ) ] link:\doibase 10.1103/physreve.85.066704 [ * * , ( ) ] http://www.amazon.com/introduction-dynamics-cambridge-mathematical-library/dp/0521663962%3fsubscriptionid%3d1v7vtj4ha4mft9xbj1r2%26tag%3dmekentosjcom-20%26linkcode%3dxm2%26camp%3d2025%26creative%3d165953%26creativeasin%3d0521663962[__ ] , ed .
( , , ) http://www.amazon.com/classical-dynamics-a-contemporary-approach/dp/0521636361%3fsubscriptionid%3d1v7vtj4ha4mft9xbj1r2%26tag%3dmekentosjcom-20%26linkcode%3dxm2%26camp%3d2025%26creative%3d165953%26creativeasin%3d0521636361[__ ] , ed .
( , , ) link:\doibase 10.1143/jpsj.77.074007 [ * * , ( ) ] https://acces-distant.upmc.fr/http//pre.aps.org/abstract/pre/v71/i3/e036707 [ * * , ( ) ] link:\doibase 10.1140/epje / i2007 - 10332-y [ * * , ( ) ] http://www.amazon.com/microhydrodynamics-principles-applications-sangtae-kim/dp/0486442195%3fsubscriptionid%3d1v7vtj4ha4mft9xbj1r2%26tag%3dmekentosjcom-20%26linkcode%3dxm2%26camp%3d2025%26creative%3d165953%26creativeasin%3d0486442195[__ ] , ed .
( , , ) http://books.google.com/books?hl=en&lr=&id=mmartf5sj9oc&oi=fnd&pg=pp2&dq=dhont+introduction+to+dynamics+of+colloids&ots=vsma887cs_&sig=s9phghvl4kfvwmkuu86linnuvzy[__ ] , ed .
( , , ) http://journals.cambridge.org/production/action/cjogetfulltext?fulltextid=376906 [ * * , ( ) ] http://link.aip.org/link/?pfldas/28/2033/1 [ * * , ( ) ] link:\doibase 10.1063/1.866120 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/001046559500029f [ * * , ( ) ] link:\doibase 10.1063/1.868626 [ * * , ( ) ] http://journals.cambridge.org/production/action/cjogetfulltext?fulltextid=393344 [ * * , ( ) ] http://www.kona.or.jp/search/15_202.pdf [ * * , ( ) ] link:\doibase 10.1063/1.451199 [ * * , ( ) ] | an improved formulation of the `` smoothed profile '' method is introduced to perform direct numerical simulations of arbitrary rigid body dispersions in a newtonian host solvent .
previous implementations of the method were restricted to spherical particles , severely limiting the types of systems that could be studied .
the validity of the method is carefully examined by computing the friction / mobility tensors for a wide variety of geometries and comparing them to reference values obtained from accurate solutions to the stokes - equation . |
rankings of the elements of a set is a common daily decision making activity , such as , voting for a political candidate , choosing a consumer product , etc .
so there is a huge literature concerning the analysis and interpretation of preference rankings data .
however , if we trace back in time , we find that de borda ( 1781 ) was the first author , who outlined a simple well thought method based on a solid argument .
borda , as a member of the french academy of sciences , criticised the plurality method of choosing a new academy member and suggested , what is known as the borda count(bc ) rule , to fully rank order ( seriate ) the @xmath0 candidates based on the preferences of the @xmath1 judges .
bc has generated a large litterature , and this paper makes full use of it as much needed .
let @xmath2 denote a set of @xmath0 alternatives / candidates / items , and @xmath3 a set of @xmath1 voters / individuals / judges . in this paper
we consider the linear orderings / rankings / preferences , in which all @xmath0 objects are rank - ordered according to their levels of desirability by the @xmath1 voters .
we denote a linear order by a sequence @xmath4 , where @xmath5 means that the alternative @xmath6 is preferred to the alternative @xmath7 let @xmath8 be the set of all linear orders on @xmath9 the cardinality of @xmath8 is @xmath10 .
a _ voting profile _ is a function @xmath11 from @xmath3 to @xmath8 , that is , @xmath12 we denote by @xmath13 the set of permutations of the elements of the set @xmath14 the borda score is a function @xmath15 from @xmath8 to @xmath16 where for a linear ordering @xmath17 @xmath18 , borda assigned to the element @xmath19the score of @xmath20 , because @xmath19 is preferred to ( @xmath21 other alternatives , or equivalently it is the @xmath22th most preferred alternative . we denote @xmath23 , where @xmath24 is a matrix having @xmath1 rows and @xmath25 columns , and @xmath26 designates the borda score of the @xmath27th judge s preference of the @xmath22th alternative .
the average borda score of the elements of @xmath28 is @xmath29 where @xmath30 is a column vector of @xmath31 s having @xmath1 coordinates .
borda s count rule ( bc ) seriates / orders the @xmath0 elements of the set @xmath28 according to their average scores : @xmath32 means alternative @xmath22 is preferred to alternative @xmath27 .
we define the reverse borda score to be a function @xmath33 from @xmath18 to @xmath16 where for a linear order @xmath17 @xmath34 , we assign to the element @xmath19the score of @xmath35 we denote @xmath36 , where @xmath37 is a matrix having @xmath1 rows and @xmath0 columns , and @xmath38 designates the reverse borda score of the @xmath27th judge s preference of the @xmath22th alternative .
the average reverse borda score of the elements of @xmath28 is @xmath39 we note that @xmath40and @xmath41 this example has two aims : first to make the notation clear ; second to show that traditional well established methods for rank data , such as distance and latent class based , may mask groups of small sizes in mixture models .
table 1 introduces a well known data set first analyzed by croon ( 1989 ) ; the data derive from a german survey of 2262 rankings of four political items concerning inglehart ( 1977 ) s theory of postmodernism .
the four items are : ( @xmath42 ) maintaining order in the nation ; ( @xmath43 ) giving people more to say in important government decisions ; ( @xmath44 ) fighting rising prices ; ( @xmath45 ) protecting freedom of speech .
inglehart advanced the thesis that there is a shift in political culture in europe ; that is , some younger europeans have different political values than their fathers : he named the elder europeans as materialists , because after the first and second world wars , they valued mostly material security item ( @xmath44 ) and domestic order item ( @xmath28 ) ; while he named some of the younger generation as postmaterialists , because they valued much more human rights and political liberties item ( @xmath45 ) and democracy item ( @xmath43 ) .
so in this example @xmath46 , @xmath47 , and the voting profile is displayed in the first two columns of table 1 ; similarly , table 1 displays the borda scores @xmath48 and the reverse borda scores @xmath49 .
the average bc score @xmath50 and the average reverse bc score @xmath51 show that , the 2262 voters generally rank materialist items @xmath52 postmaterialist items @xmath53 . + item & observed & & + ordering & & & b & c & d & & b & c & d + a@xmath54b@xmath54c@xmath54d & & & 2 & 1 & 0 & & 1 & 2 & 3 + a@xmath54b@xmath54d@xmath54c & & & 2 & 0 & 1 & & 1 & 3 & 2 + a@xmath54c@xmath54b@xmath54d & & & 1 & 2 & 0 & & 2 & 1 & 3 + a@xmath54c@xmath54d@xmath54b & & & 0 & 2 & 1 & & 3 & 1 & 2 + a@xmath54d@xmath54b@xmath54c & & & 1 & 0 & 2 & & 2 & 3 & 1 + a@xmath54d@xmath54c@xmath54b & & & 0 & 1 & 2 & & 3 & 2 & 1 + b@xmath54a@xmath54c@xmath54d & & & 3 & 1 & 0 & & 0 & 2 & 3 + b@xmath54a@xmath54d@xmath54c & & & 3 & 0 & 1 & & 0 & 3 & 2 + b@xmath54c@xmath54a@xmath54d & & & 3 & 2 & 0 & & 0 & 1 & 3 + b@xmath54c@xmath54d@xmath54a & & & 3 & 2 & 1 & & 0 & 1 & 2 + b@xmath54d@xmath54a@xmath54c & & & 3 & 0 & 2 & & 0 & 3 & 1 + b@xmath54d@xmath54c@xmath54a & & & 3 & 1 & 2 & & 0 & 2 & 1 + c@xmath54a@xmath54b@xmath54d & & & 1 & 3 & 0 & & 2 & 0 & 3 + c@xmath54a@xmath54d@xmath54b & & & 0 & 3 & 1 & & 3 & 0 & 2 + c@xmath54b@xmath54a@xmath54d & & & 2 & 3 & 0 & & 1 & 0 & 3 + c@xmath54b@xmath54d@xmath54a & & & 2 & 3 & 1 & & 1 & 0 & 2 + c@xmath54d@xmath54a@xmath54b & & & 0 & 3 & 2 & & 3 & 0 & 1 + c@xmath54d@xmath54b@xmath54a & & & 1 & 3 & 2 & & 2 & 0 & 1 + d@xmath54a@xmath54b@xmath54c & & & 1 & 0 & 3 & & 2 & 3 & 0 + d@xmath54a@xmath54c@xmath54b & & & 0 & 1 & 3 & & 3 & 2 & 0 + d@xmath54b@xmath54a@xmath54c & & & 2 & 0 & 3 & & 1 & 3 & 0 + d@xmath54b@xmath54c@xmath54a & & & 2 & 1 & 3 & & 1 & 2 & 0 + d@xmath54c@xmath54a@xmath54b & & & 0 & 2 & 3 & & 3 & 1 & 0 + d@xmath54c@xmath54b@xmath54a & & & 1 & 2 & 3 & & 2 & 1 & 0 + @xmath50 & & & 1.10 & 2.05 & 0.88 & & & & + @xmath51 & & & & & & & 1.90 & 0.95 & 2.12 + table 2 provides a statistical summary of four methods of data analysis of table 1 .
the first method suggested by inglehart is _ deductive and supervised _ ; it opposes to the other three methods , which are _ inductive , unsupervised and aim to validate inglehart s theory of postmaterialism , _ see also moors and vermunt ( 2007 ) .
the other three methods are mixture models and they attempt to see if this data set confirms inglehart s theory of postmaterialism .
the first one is by croon ( 1989 ) , who used a stochastic utility ( su ) based latent class model ; the second one by lee and yu ( 2012 ) , who used a weighted distance - based footrule mixture model ; and the third one is based on taxicab correspondence analysis ( tca ) , which is the topic of this paper . here , we provide some details on the statistics displayed in table 2 .
\a ) inglehart ( 1977 ) _ apriori _ classified the respondents into three groups : materialists , postmaterialists and mixed .
his method of classification is based on partial rankings based on the first two preferred choices . here , we discuss each group separately .
materialists are defined by their response patterns @xmath55
@xmath56 @xmath57 @xmath58 , where the pair of materialist items @xmath59 are always ranked above the pair of postmaterialist items @xmath60 ; they make @xmath61 of the voters . in the ideal case
we expect to have the average bc scores for the four items to be : @xmath62 and @xmath63 the corresponding observed values , displayed in table 2 , are ( very near to the ideal ones ) : @xmath64 and @xmath65 postmaterialists are defined by response paterns ( @xmath66 @xmath67 @xmath68 @xmath69 ) , where the pair of postmaterialist items @xmath60 are always ranked above the pair of materialist items @xmath59 ; they make @xmath70 of the voters .
the comparison of ideal and observed average bc scores , displayed in table 2 , show that : @xmath71 is very near to @xmath72 while @xmath73 is somewhat near to @xmath74 @xmath75 .
the last group is named mixed by inglehart and is composed of the remaining sixteen response patterns ; they make @xmath76 of the voters . in the ideal case
we expect to have the average bc scores for the four items to be : @xmath77 @xmath78 the corresponding observed values , displayed in table 2 , are ( somewhat near to the ideal ones ) : @xmath79 @xmath80 @xmath81 and @xmath82 furthermore , based on the global homogeneity coefficient ghc in % : @xmath83 and @xmath84 inglehart s mixed group is not globally homogenous ; that is why we did not calculate its ghc index .
the development of the ghc index and its interpretation will be done in section 3 .
it is important to note that , the underlying hypothetical conceptual - structural model for this data is a mixture composed of three specific groups ( materialist , postmaterialist and mixed ) , which are explicitly characterized by inglehart .
b , c ) given that , croon s su model and lee and yu s weighted distance - based footrule mixture model produced globally very similar groups , we present them together .
a summary of croon s analysis can also be found in skrondal and rabe - hesketh ( 2004 , p.404 - 406 ) , lee and yu ( 2012 ) and in alvo and yu ( 2014 , p.228 - 232 ) .
sections b and c of table 2 are taken from alvo and yu ( 2014 , p. 230 ) , who present a summary and a comparison of results from croon ( 1989 ) and lee and yu ( 2012 ) .
the interpretation of the estimated parameters of the su model in table 2 is similar to the average borda score : for each group the score @xmath85 shows the intensity of the preference for that item in an increasing order .
there are two kinds of estimated parameters in lee and yu s weighted distance - based footrule mixture model : the modal response pattern for each group is shown in the last column ; and the weight of an item @xmath86 , which reflects our confidence in the ranked position of the item in the modal response pattern , the higher value representing higher confidence .
both methods find a mixture of three groups similar in contents : the first two groups represent materialists with @xmath87 and @xmath88 of the voters for the weighted footrule mixture model , and @xmath89 and @xmath90 of the voters for the su mixture model ; and the third group represents postmaterialists with @xmath91 of the voters for the weighted footrule mixture model , and @xmath92 for the su mixture model .
lee and yu ( 2012 ) s conclusion is : _ based on our grouping , we may conclude that inglehart s theory is not appropriate in germany_. this assertion shows that , the two well established traditional methods masked the existence of the mixed group as put forth by inglehart .
\d ) our approach , based on taxicab correspondence analysis ( tca ) , which is an l@xmath93 variant of correspondence analysis ( ca ) , discovers a mixture of three globally homogenous groups as advocated by inglehart : materialists with @xmath94 of the voters , postmaterialists with @xmath95 , and mixed with @xmath70 of the voters .
furthermore , there is an outlier response pattern ( @xmath96 ) representing @xmath97 of the voters .
so contrary to lee and yu ( 2012 ) s assertion , _ our results validates inglehart s theory of postmodernism for this data set_. probably , this is due mainly to the fact that tca is a directional method specially useful for spherical data : rank data with all its permutations is spherical by nature ( graphically , it is represented by a permutahedron ; see marden ( 2005 , figure 2.4 , page 11 ) or benzcri ( 1980 , p.303 ) .
furthermore , based on the global homogeneity coefficient ghc in % : @xmath98 , @xmath99 and @xmath100 we see that the materialist voters form much more globally homogenous group than the voters in the mixed group ; and the voters in mixed group are much more homogenous than the voters in the postmaterialist group . furthermore , our analysis clearly shows why the postmaterialists ( they have three poles of attractions as defined by marden ( 1995 , ch . 2 ) or benzcri ( 1966 , 1980 ) ) are much more heterogenous than the materialists ( they have two poles of attractions ) .
more details on the local heterogeneities of each group will be presented later on in section 4 .
+ + group & sample% & @xmath101 & @xmath102 & @xmath103 & @xmath104 & + materialist & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & + postmaterialist & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & + mixed & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & + + group & sample% & @xmath120 & @xmath121 & @xmath122 & @xmath123 & + materialist 1 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & @xmath128 & + materialist 2 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & + postmaterialist & @xmath134 & @xmath135 & @xmath136 & @xmath137 & @xmath138 & + + group & sample% & @xmath139 & @xmath140 & @xmath141 & & + materialist 1 & @xmath142 & @xmath143 & @xmath144 & @xmath145 & & + materialist 2 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & + post - materialist & @xmath151 & @xmath152 & @xmath145 & @xmath153 & & + + group & sample% & @xmath101 & @xmath102 & @xmath103 & @xmath104 & + materialist & @xmath154 & @xmath155 & @xmath156 & @xmath157 & @xmath158 & + postmaterialist & @xmath159 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & + mixed & @xmath110 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & + outlier & @xmath168 & @xmath169 & @xmath145 & @xmath31 & @xmath170 & + the traditional methods of finding mixture components of rank data are mostly based on distance and latent class models ; these models may mask groups of small sizes ; probably due to the spherical nature of rank data . in this paper ,
our apparoach diverges from the traditional methods , because we discuss the concept of a mixture for rank data essentially in terms of its globally homogenous group components .
we use the law of contradiction to identify globally homogenous components .
for instance , by tca we were able to discover that the data set in table 1 is a mixture of three globally homogenous group components ( materialist , postmaterialist and mixed ) ; furthermore , each group component can be summarized by its average borda count ( bc ) score as its consensus ranking ; this is the first step in our procedure . in the second step , we look at local heterogeneities if there are any , given the globally homogenous component .
this two step procedure produces finer visualization of rank data ; it is done via the exploratory analysis of rank data by taxicab correspondence analysis with the nega coding .
also we introduce a new coefficient of global homogeneity , ghc .
ghc is based on the first taxicab dispersion measure : it takes values between 0 and 100% , so it is easily interpretable .
ghc measures the extent of crossing of scores of voters between 2 or 3 blocks seriation of the items where the borda count statistic provides consensus ordering of the items on the first axis .
furthermore , to our knowledge , this is the first time that a tangible method has been proposed that identifies explicitly outliers in a rank data : neither the recently written monograph by alvo and yu ( 2014 ) , nor the much cited monograph of marden ( 1995 ) discuss the important problem of identification of outliers in rank data .
we mention two publicly available written packages in r , that we used , _ rankclustr _ by jacques , grimonprez and biernacki ( 2014 ) , and _ pmr _ ( probability models for ranking data ) by lee and yu ( 2013 ) . the contents of this paper are organized as follows : section 2 reviews the tca approach for rank data ; section 3 develops the new global homogeneity coefficient ghc ; section 4 presents the analysis of some well known rank data sets by tca ; and finally in section 5 we conclude with some remarks .
we just want to mention that there is a large litterature in social choice theory or social welfare theory studying the properties of the bc . here , we mention some important contributions according to our personal readings .
young ( 1974 ) presents a set of four axioms that characterize uniquely bc ; see also among others , saary ( 1990a ) and marchant ( 1998 ) .
saari ( 1990b ) distinguishes two levels of susceptiblity of manipulation of voting theories : _macro_- where a large percentage of voters- , and _micro_- where a small percentage of voters - attempt to change the results of the elections . in data analysis ,
a macro manipulation is equivalent to the existence of a mixture of groups of voters . while , a micro manipulation is equivalent to the existence of few outliers in the globally homogenous set of voters @xmath3 .
further , saari concludes that among all positional voting systems , bc is the least susceptible to micro manipulation ; this assertion seems fully true in this paper .
saari ( 1999 ) proves that bc is the only positional voting method that satisfies the property of _ reversal symmetry , _ which states that if everyone reverses all their preferences , then the final outcome should also be reversed .
this property plays an important role in the nega coding of a rank data set before the application of tca .
choulakian ( 2014 ) incorporates the bc to interpret the first principal factor of taxicab correspondence analysis ( tca ) of a nega coded rank data , see theorem 1 in the next section .
additionally , this essay further extends and complements the ideas of global homogeneity and local heterogeneities for rank data .
results of this section are taken from choulakian ( 2006 , 2014 ) .
we start with an overview of tca of a contingency table ; then review the corresponding results concerning rank data .
let @xmath171 be a contingency table cross - classifying two nominal variables with @xmath172 rows and @xmath173 columns , and @xmath174 be the associated correspondence matrix with elements @xmath175 where @xmath176 is the sample size .
we define as usual @xmath177 , @xmath178 the vector @xmath179 the vector @xmath180 , and @xmath181 a diagonal matrix having diagonal elements @xmath182 and similarly @xmath183 let @xmath184 in tca we compute the following quadruplets for the two spaces , for @xmath185 : @xmath186 in the row space of @xmath187 and @xmath188 in the column space of @xmath189 . given that in ca and tca , the row and column spaces are dual to each other , we name the pair of vectors ( @xmath190 and @xmath191 @xmath192th principal axes , the pair ( @xmath193 and @xmath194 @xmath192th basic vectors of coordinates , the pair ( @xmath195 and @xmath196 @xmath192th vectors containg tca factor scores , and the nonnegative scalar @xmath197 the @xmath192th tca dispersion measure . the relations among the seven terms will be described in the next two subsections .
tca is computed in 2 steps : in the first step we compute the taxicab singular value decomposition ( tsvd ) of @xmath187 as a function of @xmath198 for @xmath199 which is a stepwise matrix decomposition method based on a particular matrix norm , see below equation ( 3 ) . in the 2nd step , we reweight the pair of basic vectors @xmath200 by respective weights of the columns , @xmath201 and the rows , @xmath202 to obtain the vectors of factor scores @xmath203 for @xmath204 .
let @xmath205 be the residual data matrix at the @xmath192th iteration , where , @xmath206 for @xmath207 tsvd consists of maximizing the @xmath208 norm of the linear combination of the columns of the matrix @xmath205 subject to @xmath209 norm constraint , where the @xmath208 norm of a vector @xmath210 is defined to be @xmath211 @xmath212and @xmath213 is the @xmath209 norm ; more precisely , it is based on the following optimization problem@xmath214or equivalently , it can also be described as maximization of the @xmath208 norm of the linear combination of the rows of the matrix @xmath215@xmath216equation ( 1 ) is the dual of ( 2 ) , and they can be reexpressed as matrix operator norms @xmath217which is a well known and much discussed matrix norm related to the grothendieck problem ; the inequality in theorem 2 section 3 of this paper sheds further insight into grothendieck s theorem ; see pisier ( 2012 ) for a comprehensive and interesting history of grothendieck s theorem with its many variants .
equation ( 3 ) characterizes the robustness of the method , in the sense that , the weights affected to the columns ( similarly to the rows by duality ) are uniform @xmath218 the @xmath192th principal axes , @xmath190 and @xmath219 are computed by@xmath220and@xmath221it is evident that for @xmath222@xmath223where @xmath224 represents a column vector of ones of length @xmath225 the two principal axes @xmath226 and @xmath227 are named trivial , and they are used only to center the rows and the columns of @xmath187 . let @xmath193 represent the @xmath192th tsvd coordinates of the rows of @xmath205 by projecting the rows of @xmath205 on the principal axis @xmath190 , and @xmath228 represent the @xmath192th tsvd coordinates of the columns of @xmath205 by projecting the columns of @xmath205 on the principal axis @xmath229 .
these are given by @xmath230 and@xmath231 in particular , by ( 6,7,8 ) , we have for @xmath222@xmath232equations ( 7 ) are named transition formulas , because @xmath233 and @xmath234 and , @xmath190 and @xmath235 are related by @xmath236where @xmath237 and @xmath238 if @xmath239 @xmath240 otherwise . to obtain the @xmath241th tsvd row and column coordinates @xmath242 and @xmath243 and corresponding principal axes @xmath244 and
@xmath245 , we repeat the above procedure on the residual dataset @xmath246we note that the @xmath247 because by ( 6 ) through ( 9 ) @xmath248which , by induction , implies that for @xmath204 @xmath249and in particular we see that for @xmath204 @xmath250that is , the basic vectors @xmath228 and @xmath193 are centered .
the data reconstitution formula for the correspondence matrix @xmath187 as a function of the basic vectors @xmath200 for @xmath251associated with the dispersion measures @xmath252 is@xmath253 in tca of @xmath187 both basic vectors @xmath193 and @xmath254 for@xmath255 satisfy the equivariability property , as a consequence of equations ( 8,12 ) , see choulakian ( 2008a ) .
this means that @xmath193 and @xmath228 are balanced in the sense that@xmath256 \nonumber \\ & = & -\sum_{i}\left [ a_{\alpha } ( i)|a_{\alpha } ( i)<0\right ] \\ & = & \sum_{j}\left [ b_{\alpha } ( j)|b_{\alpha } ( j)>0\right ] \nonumber \\ & = & -\sum_{j}\left [ b_{\alpha } ( j)|b_{\alpha } ( j)<0\right ] .
\nonumber\end{aligned}\ ] ] in tsvd , the optimization problems ( 3 ) , ( 4 ) or ( 5 ) can be accomplished by two algorithms .
the first one is based on complete enumeration ( 3 ) ; this can be applied , with the present state of desktop computing power , say , when @xmath257 the second one is based on iterating the transitional formulas ( 7 ) , ( 8) and ( 9 ) , similar to wold s ( 1966 ) nipals ( nonlinear iterative partial alternating least squares ) algorithm , also named criss - cross regression by gabriel and zamir ( 1979 ) .
the criss - cross nonlinear algorithm can be summarized in the following way , where @xmath258 is a starting value : step 1 : @xmath259 , @xmath260 and @xmath261 step 2 : @xmath262 @xmath263 and @xmath264 step 3 : if @xmath265 go to step 1 ; otherwise , stop .
this is an ascent algorithm , see choulakian ( 2016 ) ; that is , it increases the value of the objective function @xmath266 at each iteration .
the convergence of the algorithm is superlinear ( very fast , at most two or three iterations ) ; however it could converge to a local maximum ; so we restart the algorithm @xmath172 times using each row of @xmath205 as a starting value .
the iterative algorithm is statistically consistent in the sense that as the sample size increases there will be some observations in the direction of the principal axes , so the algorithm will find the optimal solution .
a simple reweighting of the basic coordinates @xmath267 for@xmath255 produces the vectors @xmath268 that contain tca factor scores of the rows and the columns @xmath269 and ( 8) becomes@xmath270 similar to ca , tca satisfies an important invariance property : columns ( or rows ) with identical profiles ( conditional probabilities ) receive identical factor scores . moreover , merging of identical profiles does not change the result of the data analysis : this is named the principle of equivalent partitioning by nishisato ( 1984 ) ; it includes the famous distributional equivalence property of benzcri , which is satisfied by ca . by ( 13 and 15 ) , one gets the data reconstitution formula in tca ( the same formula as in ca ) for the correspondence matrix @xmath187 as a function of the factor coordinates @xmath203 for @xmath271associated with the eigenvalues @xmath197 @xmath272 .\]]the visual maps are obtained by plotting the points @xmath273 for @xmath274 or @xmath275 for @xmath276 for @xmath277 correspondence analysis ( ca ) admits a chi - square distance interpretation between profiles ; there is no chi - square like distance in tca .
fichet ( 2009 ) described it as a general scoring method . in the sequel
we suppose that the theory of correspondence analysis ( ca ) is known .
the theory of ca can be found , among others , in benzcri ( 1973 , 1992 ) , greenacre ( 1984 ) , gifi ( 1990 ) , le roux and rouanet ( 2004 ) , murtagh ( 2005 ) , and nishisato ( 2007 ) ; the recent book , by beh and lombardi ( 2014 ) , presents a panoramic review of ca and related methods .
further results on tca can be found in choulakian et al .
( 2006 ) , choulakian ( 2008a , 2008b , 2013 ) , choulakian and de tibeiro ( 2013 ) , choulakian , allard and simonetti ( 2013 ) , choulakian , simonetti and gia ( 2014 ) , and mallet - gauthier and choulakian ( 2015 ) . in the sequel
, we use the same notation as in choulakian ( 2014 ) .
let @xmath278 for @xmath279 and @xmath280 represent the borda scores for rank data , where @xmath26 takes values @xmath281 similarly , @xmath49 represent the reverse borda scores .
we note that @xmath48 and @xmath282 contain the same information . to incorporate both in one data set , there are two approaches in correspondence analysis literature . in the first approach we vertically concatenate both tables , that is ,
we double the size of the coded data set by defining @xmath283 ... @xmath284 in the second approach , we summarize @xmath285 by its column total , that is , we create a row named @xmath286 @xmath287 then we vertically concatenate @xmath288 to @xmath48 , thus obtaining@xmath289where @xmath290 the size of @xmath291 is @xmath292 choulakian ( 2014 ) discussed the relationship between tca of @xmath293 and tca of @xmath294 we will consider only the application of tca to @xmath291 , because this will show if the rank data set is globally homogenous or heterogenous .
so let @xmath295be the correspondence table associated with @xmath294 note that @xmath296 is a matrix of size @xmath297 and @xmath298 is a row vector of size @xmath299 .
we denote the sequence of principal axes , and the associated basic vectors , tca vectors of principal factor scores and dispersion measures by @xmath300 for @xmath204 and @xmath301 .
the following theorem is fundamental , and it relates @xmath302 the average bc score of items , to the first principal tca factor score @xmath303.@xmath304 * theorem 1 ( * @xmath305 ) : * ( * this is theorem 2 in choulakian ( 2014 ) * ) * : properties a , b , c are true iff @xmath306 , where @xmath307 is the first principal axis of the columns of @xmath308 \a ) the first principal column factor score @xmath303 of the @xmath0 items is an affine function of the average bc score @xmath309 that is , @xmath310 \b ) the first nontrivial tca dispersion measure equals twice taxicab norm of the row vector @xmath311 @xmath312 \c ) consider the residual matrix @xmath313 then @xmath314 that is , the nega row @xmath315 is the null row vector .
note that in theorem 1 we have eliminated sign - indeterminacy of the first principal axis , by fixing @xmath316 property a implies that the first principal factor score of the items , @xmath317 can be interpreted as the borda ranking of the @xmath0 items .
property b shows that the nega row of @xmath318 accounts for 50% of the first nontrivial taxicab dispersion @xmath319 property c shows that the residual matrix @xmath320 does not contain any information on the heavyweight nega row .
properties b and c imply that the first nontrivial factor is completely determined by the nega row , which plays a dominant heavyweight role , see choulakian ( 2008a ) .
such a context in ca is discussed by benz**cri ( 1979 ) using asymptotic theory , and in dual scaling by nishisato ( 1984 ) , who names it forced classification .
the importance of applying tca to nega coded dataset , * r*@xmath321 , and not to the original data set * r * stems from the following two considerations : first , if there are two columns in * r * such that @xmath322 for @xmath323 and @xmath324 then the columns @xmath22 and @xmath325 have identical profiles , and by the invariance property of tca they can be merged together , which will be misleading .
second , as discussed by choulakian ( 2014 ) , the interpretation of tca maps of * r*@xmath321 is based on the law of contradiction , which will be used recursively to find the mixture components .
let @xmath326 be a statement and and @xmath327 its negation ; then the law of contradiction states that @xmath326 and @xmath327 oppose each other : they can not both hold together , see for instance eves ( 1990 ) .
we shall use the law of contradiction as a basis for the interpretation of the maps produced by tca of * r*@xmath321 in the following way .
first , we recall that there are @xmath0 items , and we represented the borda score of an item @xmath22 by the voter @xmath328 for @xmath280 and @xmath279 , by @xmath26 and its reverse borda score by @xmath329 by the law of contradiction , @xmath26 and @xmath330 oppose each other ; which in its turn also implies that @xmath26 and @xmath331 oppose each other because the @xmath332 contains @xmath333 or they are not associated at all if @xmath334 .
we let @xmath335to represent the first tca vector of factor scores of the @xmath336 rows . for the interpretation of the results by tca of * r*@xmath321 we can have the following two complementary scenarios : scenario 1 happens when@xmath337then by the law of contradiction ,
the first principal dimension is interpretable and it shows the opposition between the borda scores of the items @xmath26 to their reverse borda scores @xmath338 summarized by @xmath332 .
if scenario 1 happens , then we will say that the data set is globally homogenous , because all voters have positive first tca factor scores ; that is , they are directionally associated because @xmath339 for all @xmath340 now by property c of theorem 1 , the nega row disappears and do not contribute to the higher dimensions ; thus the higher dimensions will exhibit either random noise or local heterogeneities of the voters represented by their response patterns .
scenario 2 is the negation of scenario 1 , it corresponds to @xmath341then the results of tca of * y*@xmath321 are not interpretable by the law of contradiction : because some voters , say belonging to the subset v@xmath342 are directionally associated with the nega ; so to obtain interpretable results as described in scenario 1 , we eliminate the subset of voters v@xmath93 , and repeat the analysis till we obtain scenario 1 . if the number of deleted voters in v@xmath93 is small , we consider them outliers ; otherwise , they constitute another group(s ) of voters .
we have the following * definition 1 * : if scen1 holds , then we name the rank data * r * or * r*@xmath321 globally homogenous . _
it is of basic importance to note that using r_@xmath321 _ , only globally homogenous data are interpretable by the law of contradiction . _
rank data is much more structured than ratings data ; and this aspect will be used to propose a global homegeneity coefficient ( ghc ) of rank data .
we recall that @xmath343is the nega coded rank data and @xmath295its associated correspondence matrix .
we note the following facts : _ fact 1 _ : the row sum of the elements of @xmath291 are : @xmath344 for rows @xmath279 and @xmath345 for the nega row ( or @xmath336th row ) . from which we get the total sum of elements of @xmath291 to be @xmath346 .
so , the marginal relative frequency of the @xmath27th row is @xmath347 for @xmath279 , and , the marginal relative frequency of the nega row is @xmath348 .
fact 2 _ : the column sum of the elements of @xmath291 are : @xmath349 for columns @xmath350 so , the marginal relative frequency of the @xmath22th column is @xmath351 for @xmath352 _ fact 3 _ : using facts 1 and 2 , we see that the first residual matrix @xmath353has elements of : @xmath354\end{aligned}\]]and@xmath355equation ( 18 ) states that @xmath318 is row centered with respect to average ranking @xmath356 , because @xmath357 for @xmath358 and @xmath359 ; it is also column centered .
we have the following * proposition 1 : * for a globally homogenous rank data , @xmath360 .
the proofs of new results are in the appendix .
young ( 1974 ) presented a set of four axioms that characterize uniquely bc rule .
his axiom 4 , named _
faithfulness , _ states that when there is only one voter , if the relation that he uses to express his preferences is so simple that one result seems the only reasonable one , the result of the method must be that one .
the _ faithfulness _
axiom was the inspiration of this section . by the invariance property of tca ,
that is , merging of identical profiles does not change the results of the data analysis , a _ faithfully homogenous group _ is equivalent to the existence of one response pattern @xmath361 , and its borda score values for a complete linear order of @xmath25 items can be represented by @xmath362 , without loss of generality by reordering of the items .
then the nega coded correspondence table , @xmath363 will have only two rows and @xmath0 columns@xmath364and @xmath318 will be of rank 1 ; that is , there will be only one principle factor , for which we note its taxicab dispersion measure by @xmath365 for a fixed finite integer value of @xmath366the following result gives the value of @xmath367 explicitly .
* theorem 2 ( faithfully homogenous group ) * : \a ) for @xmath368 or @xmath369 for @xmath370 , then@xmath371where we define @xmath372 \b ) the first and only factor score of the two rows are @xmath373 @xmath374 and @xmath375 \c ) the first and only factor score of the @xmath22th item is @xmath376 for @xmath352 in particular , we see that @xmath377 @xmath378 and @xmath379 for @xmath280 are equispaced .
so for an odd number of items @xmath380 for @xmath370 , we have @xmath381 let @xmath382 denote the cardinality of a set @xmath28 , that is , the number of elements in @xmath28 .
the next definition formalizes the partition of a set of items obtained in theorem 1 . *
definition 2 * : for a globally homogenous rank data set , we define a partition of a set of items @xmath28 to be _ faithful _ if a ) for an even number of items @xmath368 and @xmath383 the first tca axis divides the set of items @xmath28 into 2 blocks such that @xmath384 where @xmath385 and @xmath386 \b ) for an odd number of items @xmath368 @xmath387 for @xmath383 the first tca axis divides the set of items @xmath28 into 3 blocks @xmath388 where @xmath389 and , @xmath390 and @xmath391 * remarks 1 * \a ) the bc for a faithfully homogenous group is @xmath392 , and the pearson correlation _
corr(_@xmath393 as in theorem 1a .
\b ) theorem 2 concerns only one group .
theorem 4 generalizes theorem 2 to multiple faithfully homogenous subgroups ; however in the multiple case only parts a and b of theorem 2 are satisfied and not part c. the maximum number of multiple faithfully homogenous subgroups is @xmath394 for @xmath368 or @xmath395 and @xmath383 which represents the number of within ( intra ) block permutations of the rankings .
the next result shows that @xmath367 is an upper bound for the first tca dispersion measure @xmath396 * theorem 3 * : for a globally homogenous rank data set we have@xmath397 * definition 3 : * based on theorem 3 we define for a globally homogenous rank data the following global homogeneity coefficient@xmath398
@xmath399 takes values between 0 and 100 . in real applications
we seldom find the value of @xmath400 .
however , it may approach 100% as in the potato s rank data set considered later on .
* theorem 4 * : @xmath400 if and only if there is a faithful partition of the items and the borda scores of all voters are intra ( within ) block permutations .
* corollary 1 * : @xmath400 if and only if @xmath401 @xmath402 for @xmath279 .
the following result complements proposition 1 .
* corollary 2 : * for a globally homogenous rank data , @xmath403 for @xmath279 . * definition 4 * : a voter is named faithful if its first factor score @xmath404 in the next subsection we explain these results . *
we consider the following artificial example with two voters and eight items .
[ cols= " < , > , > , > , > , > , > , > , > " , ] for this example we have : @xmath405 , @xmath406 and @xmath407 so @xmath408 figure 2 displays the tca biplot .
we note : first , the first axis subdivides the items into 2 faithful blocks @xmath409 and @xmath410 additionaly , the ordering of the items on the first axis is given by the borda count @xmath302 where we see that @xmath411 second , voters 1 and 2 are confounded , because they have the same profile ; further @xmath412 so voters 1 and 2 are faithful .
third , @xmath413 which is smaller in value than @xmath414 because voter 3 scores cross the two faithful blocks : score 4 has crossed the block @xmath415 to the block @xmath416 and score 3 has crossed the block @xmath417 to the block @xmath415 .
the second axis will represent local heterogeneity or random error .
* as an application of theorem 4 , we consider inglehart s a priori classification summarized in part a in table 2 .
we mentioned that , inglehart s _ materialist _ group , composed of 4 response patterns , has @xmath418 and the _ postmaterialist _ group , also composed of other 4 response patterns , has also @xmath400 ; there is no value of ghc for the _ mixed _ group , because the _ mixed _ group is not globally homogenous .
let us consider inglehart s _ materialist _ group where the pair of materialist items @xmath59 are always ranked above the pair of postmaterialist items @xmath60 ; from table 1 we have ( where we permuted the positions of items c and b):@xmath419we see that the subsets @xmath59 and @xmath60 form a faithful 2 blocks partition of the four items with no crossing between the blocks , so theorem 4 applies and @xmath420 * we consider the following four orderings found in table 1 with their frequencies : @xmath421 , @xmath422 , @xmath423 and @xmath424 where abcd137 represents the ordering @xmath425 with its frequency of 137 .
figure 3 displays the tca biplot .
we can summarize the data analysis by the following observations concerning the first axis : first , by definition 1 the rank data is globally homogenous , because the factor scores of the 4 response patterns are positive on the first axis .
second , item @xmath28 opposes to the items @xmath426 so the partition of the items is not faithful .
evidently , this means that all voters ranked item a as their first choice ; however there is considerable heterogeneity concerning the rankings of the other 3 items @xmath427 these local heterogeneities will appear on the second and third axes ( not shown ) .
third , @xmath428 for @xmath429and@xmath430 while @xmath431 by theorem 2@xmath432 so @xmath433 .
this example shows that the condition @xmath434 for @xmath279 is not sufficient for @xmath435 * the above discussion shows that : * ghc takes into account the following two aspects of rank data : a ) how the items are partitioned into blocks by the first axis .
b ) the extent of crossing of scores of voters among the partitioned blocks , where the borda count statistic provides consensus ordering of the items on the first axis*. @xmath400 means that the 2 or 3 blocks are faithful and for all voters their orderings of the items are intra block permutations with no crossing between the blocks .
* example 1 * : we consider tca of the potatos rank data set , found in vitelli et al .
( 2015 , table a2 ) ; it has @xmath436 assessors and @xmath437 potatos .
the first four tca dispersion measures are : @xmath438 , @xmath439 and @xmath440 ; by theorem 2 , @xmath441 so @xmath442 , which is very high .
figure 4 displays the tca biplot of the assessors ( as points ) and the true ranks of the potatos as provided in their paper .
first , on the first axis we observe a _
faithful _ partition of the 20 potatos into 2 blocks : ten potatos t11 to t20 are found on the left side of the first axis , and ten potatos t1 to t10 are found on the right side of the first axis .
however , potatos numbered 5 to 10 are not correctly ranked ( the true ranks of the 3 pairs @xmath443 @xmath444 , and @xmath445 are permuted ) .
the distribution of the majority of the true ranks of the potatos on the first axis seem uniform .
second : the first factor score of the assessors * * @xmath446 for @xmath447 has values @xmath448 ( 4 times ) , @xmath449 ( 2 times ) , @xmath450 ( 3 times ) and @xmath451 ( 3 times ) , which are highly clustered around @xmath452 .
these values show that only three assessors are faithful ; the other 9 assessors scores have some inter block crossings ; by examining the signs of @xmath453 , the crossings happened between the subsets of items @xmath445 and @xmath454 , which are near the origin ; and this is the reason that @xmath455 did not attain its upper value of 100% but it approached it .
third , given that the rank data is globally homogenous , the bc vector @xmath50 reflects the consensus linear ordering of the potatos on the first axis , because by theorem 1a , @xmath456 fourth , the data is almost unidimensional , because @xmath457 approaches 0 and this is apparent in figure 1 . in conclusion
, we can say that this is a nice ideal real data set almost faithfully homogenous with some random sampling error .
the notion of global homogeneity and local heterogeneity will further be explained by analyzing few real data sets .
first we provide some details concerning part d of table 2 . for the other data sets we provide just essential aspects . here
we describe the six consecutive steps for the analysis of the rank data set in table 1 .
we recall the the description of the four items : ( @xmath28 ) maintaining order in the nation ; ( @xmath43 ) giving people more to say in important government decisions ; ( @xmath44 ) fighting rising prices ; ( @xmath45 ) protecting freedom of speech .
.... step 1 : tca of the full data set ....
figure 5 displays the biplot of the complete data set , where to each response pattern its observed frequency is attached ; for instance the first response pattern in table 1 , @xmath425 with observed frequency of 137 , is labeled as abcd137 in the biplot . by the law of contradiction figure 4
is not interpretable , because there are 16 response patterns associated with nega ; we recall that the point nega contains the negations of all response patterns .
note that the 8 response patterns @xmath458 that appear on the second axis have very small negative values on the first axis .
so we eliminate these 16 response patterns , which have negative first factor scores , and apply tca to the remaining 8 response patterns in step 2 .
.... step 2 : tca of the subset composed of 8 response patterns .... the application of tca to the nega coded subset of weighted 8 response patterns produces the following tca dispersion values : @xmath459 @xmath460 and @xmath461 .
figures 6 and 7 summarize the data .
figure 6 represents the biplot of the principal plane of the 8 response patterns ; it has very clear interpretation .
\a ) the first factor opposes the nega row to the 8 response patterns which represent the _ materialists _ @xmath462 : the 8 response patterns form a globally homogenous group of voters , and they represent @xmath463of the voters in the sample ; further they can be summarized by their average bc score , @xmath464 because by theorem 1a , @xmath465 note that @xmath303 contains the first factor coordinates of the four items plotted in figure 6 .
@xmath466 by theorem 2 , so the global homogeneity coefficient of this group is @xmath467 , which is relatively high . on the first axis we observe the faithful partition of the items into 2 blocks , @xmath468 and @xmath60 ; but the first factor scores of the voters have two values , @xmath469 and @xmath470 .
this implies that the response patterns cadb294 , cabd330 , acdb255 and acbd309 are faithful ; while there are inter block crossings of scores of the response patterns cdab70 , cbad117 , adcb93 and abcd137 .
this last assertion is evident .
\b ) the nega point contributes only to the first axis ; and by theorem 1c , it is eliminated from the higher axes : in figure 7 it is found at the origin .
\c ) given that the 2nd and 3rd tca dispersion values are almost equal and relatively high , @xmath460 and @xmath461 , it is worthwhile to examine the principle plane made up of axes 2 and 3 , represented in figure 7 : it is evident that there are two principle branches dominated by items a and c respectively ; these two branches represent _ local _ _ heterogeneities _ , in the sense that item c opposes to item a on both axes , which are both qualified as materialist items .
the two groups postmaterialist1 and postmaterialist2 in table 2 , which appeared as individual groups in croon s su mixture model and lee and yu s weighted distance based model , are the two local branches ( subdivisions ) of the materialists in the tca approach .
these two branches are similar to marden ( 1995 , chapter 2 ) s the poles of attraction for items a and c. .... step 3 : tca of the 16 response patterns .... we apply tca to the 16 remaining response patterns that were associated with the nega point in figure 5 , and we get figure 8 , which by the law of contradiction is not interpretable : so we eliminate the 6 response patterns , which are associated with the nega point in figure 8 ; and apply tca to the remaining 10 response patterns in step 4 ( step 4 is similar to step 2 ) .
.... step 4 : tca of the 10 response patterns .... tca dispersion measures for this case are : @xmath471 @xmath472 and @xmath473 .
figures 9 and 10 totally reflect the data .
figure 9 has the following interpretation : a ) the first factor represents the _ postmaterialists _ with _ _ @xmath474 which is low for the following two reasons : the partition of the four items into two blocks , @xmath28 and @xmath475 is unfaithful by the first axis and there are a lot of inter - block crossings by the response patterns @xmath476 .
\b ) in figure 10 , the points nega and item a are found on the origin : they do not contribute to principal axes 2 and 3 .
\c ) in figure 10 there are three principle branches dominated by items b , d and c , respectively ; these three branches represent _ local _ _ heterogeneities _ , but the two branches starting with b ( giving people more to say in important government decisions ) and d ( protecting freedom of speech ) are more important than the smaller branch starting with item c ( fighting rising prices ) . .... step 5 : tca of the 6 response patterns .... figure 11 represents the tca map of the last six patterns deleted in step 4 . in this plot
, we identify the response pattern dacb30 as an outlier because its proportion is very small @xmath477 ; so we eliminate it .
.... step 6 : tca of the remaining 5 response patterns .... tca dispersion measures are : @xmath478 and @xmath479 . in figure 12
the first factor represents the _
mixed _ group .
note that on the first axis , the mixed items @xmath480 oppose to the mixed items @xmath481 ; similarly , on the second axis , the mixed items @xmath482 oppose to the mixed items @xmath483 . here , we continue the analysis of croon s political goals data by reducing it to the first two choices , that is considering only partial rankings , as done by inglehart .
inglehart s approach is based on the first two choices of the four items ; thus the 24 response patterns of table 1 is reduced to 12 response patterns .
for example , the first two response patterns @xmath484 and @xmath485 with respective frequencies 137 and 29 , are collapsed into one response pattern @xmath486 with frequency of 166 , where * represents either c or d. now the borda score of @xmath487 @xmath488 that is the items c and d take equal scores .
the tca of the partial ranking table produces only two groups : materialists @xmath489 with two branches ( poles of attraction ) and postmaterialists @xmath490 with two branches .
figures 13 through 16 display these results . on these figures
the label , for instance cb186 , represents the partial order @xmath491 with its frequency of 186 .
it is obvious that there is loss of information by reducing the complete rankings into partial rankings .
roskam preference data of size 39 by 9 was analyzed by de leeuw ( 2006 ) and de leeuw and mair ( 2009 ) and can be downloaded from their package homals in r. in 1968 , roskam collected preference data where 39 psychologists ranked all 9 areas of the psychology department at the university of nijmengen in the netherlands .
the areas are : soc = social psychology , edu = educational and developmental psychology , cli = clinical psychology , mat = mathematical psychology and psychological statistics , exp = experimental psychology , cul = cultural psychology and psychology of religion , ind = industrial psychology , tst = test construction and validation , and lastly phy = physiological and animal psychology .
de leeuw ( 2006 ) compared linear and nonlinear principal components analysis ( pca ) approaches ( figures 4.6 and 4.7 in his paper ) , and concluded that the grouping in the nonlinear pca is clearer : psychologists in the same area are generally close together , and there is relatively clear distinction between qualitative and quantitative areas .
this assertion is true , because it describes the two component groups of the mixture identified by tca as will be seen .
later on , de leeuw and mair ( 2009 ) applied multiple correspondence analysis , named also homogeneity analysis , to this data set with the scale level ordinal , and interpreted the obtained figure ( figure 8 in their paper ) with the following conclusion : the plot shows interesting rating twins of departmental areas : @xmath492,@xmath493 , @xmath494,@xmath495 .
@xmath496 is somewhat separated from the other areas .
this assertion does not seem to be completely true , because of masking phenomenon .
our tca analysis reveals that the 39 psychologists represent a mixture of two globally homogenous groups of sizes 23 and 16 as shown in figures 17 and 18 . in figure 17 , the following borda ordering of the areas can be discerned visually : @xmath497 the quantitative areas of psychology are preferred for this group .
_ note that _ @xmath498 _ is not separated from the rest , it has a middle ranking_. in figure 18 , the following borda ordering of the areas can be discerned visually : @xmath499 for this group of psychologists phy is the worse rated area ; further , qualitative areas are preferred for this group . for the complete data set , @xmath500 , but
the resulting tca map is not interpretable . for group 1 , @xmath501 , @xmath502 , so @xmath503 ; for group 2 , @xmath504 so @xmath505 so , group1 is somehat more globally homogenous than group 2 ; however both groups have a lot of inter block crossings .
we also note that : @xmath506 and @xmath507 ; that is , the first tca dispersion measure of noninterpretable data is much smaller than the corresponding value of an interpretable maximal subset .
this data set of preferences can be found in takane ( 2014 , p.184 - 5 ) : in 1978 delbeke asked 82 belgian university students to rank - order 16 different family compositions , where the 16 orders are described by the coordinate pairs @xmath508 for @xmath509 and , the index @xmath27 represents the number of daughters and the index @xmath22 the number of sons .
this data set has been analyzed by , among others , heiser and de leeuw ( 1981 ) , van deun , heiser and delbeke ( 2007 ) , takane , kiers and de leeuw ( 1995 ) . in these studies ,
the family composition ( 0,0 ) is considered an outlier because of its high influence and sometimes omitted from analysis . in our approach
there are no outlier items , but voters can be tagged as outliers only by the law of contradiction .
our results differ from theirs : we get a mixture of two globally homogenous groups of sizes 68 and 14 as shown in figures 19 and 20 , where points represent students and the symbol @xmath510 represents the family composition @xmath508 for @xmath511 . in figure 19
, we see that for this majority group of 68 students the least preferred combination of kids is ( 0,0 ) and the most preferred combination is ( 2,2 ) .
the first borda axis opposes the combinations composed of ( 0 daughters or 0 sons ) to the combinations composed of ( at least one daughter or at least one son ) .
further , we see that there is a bias towards boys : on the first axis the position of the point @xmath508 is always to the left of the point @xmath512 . for group 1 , @xmath513 , @xmath514 , so @xmath515 ; looking at the values of the students first factor scores , we notice 2 clusters : cluster 1 , characterized by small number of inter blocks crossings , is composed of 26 students with first factor score of @xmath516 8 students with @xmath517 and 5 students with @xmath518 this cluster , of proportion @xmath519 is represented by the dots making a vertical line in figure 15 .
cluster 2 , characterized by large number of inter block crossings , are quite dispersed , having first factor scores between 0.4583 and 0.0750 . in figure 20 , for the minority group of 14 students ( labeled on the biplot by their row numbers ) the least preferred combination of kids
is @xmath520 and the most preferred combination is @xmath521 .
the first borda axis opposes the combinations composed of @xmath522 and @xmath508 such that @xmath523 to the rest .
further , in this group also there is a bias towards boys : on the first axis the position of the point @xmath508 is always to the left of the position of the point @xmath512 .
for group 2 , @xmath524 and @xmath525
here we conclude with a summary of some aspects of tca of nega coded rank data .
we note that the rank data is spherical by nature , they are represented on a permutahedron ; so a directional method , like tca of the nega coded rank data , is able to discover some other aspects of the data , which are eclipsed or masked by well established methods , such as distance or latent class based methods . like occam s razor ,
step by step , tca peels the essential structural layers ( globally homogenous groups ) of rank data ; it can also identify outliers in a group .
we presented a new coefficient , @xmath399 , that measures the global homogeneity of a group .
ghc is based on the first taxicab dispersion measure : it takes values between 0 and 100% , so it is easily interpretable .
ghc takes into account the following two aspects of rank data : a ) how the items are partitioned into blocks by the first axis .
b ) the extent of crossing of scores of voters among the partitioned blocks , where the borda count statistic provides consensus ordering of the items on the first axis .
@xmath418 means that the partition of the set of items into 2 or 3 blocks is faithful and for all voters their orderings of the items are intra block permutations with no crossings between the blocks . for fully ranked data ,
the lower bound of @xmath399 is positive but unknown being an open problem .
as is well known , a coefficient in itself does not show important local details in a data set .
we named these important local details , local heterogeneity ; and they appear in the higher dimensions of tca outputs : so it is important to examine the sequence of tca dispersion measures and the graphical displays as expounded and professed by benzcri .
benzcri , j.p .
geometric representation of preferences and correspondence tables . in _
pratique de lanalyse des donnes _ , vol .
2 , by bastin , ch . , benzcri , j.p . , bourgarit , ch . and cazes , p. p : 299 - 305 , dunod , paris . choulakian v. ( 2013 ) . the simple sum score statistic in taxicab correspondence analysis . in _ advances in latent variables _
( ebook ) , eds .
brentari e.and carpita m. , vita e pensiero , milan , italy , isbn 978 88 343 2556 8 , 6 pages .
de leeuw , j ( 2006 ) .
nonlinear principal component analysis and related techniques . in mj
greenacre , j blasius ( eds . ) , _ multiple correspondence analysis and related methods _ ,
chapter 4 , 107134 , chapman & hall / crc , boca raton .
van deun , k. , heiser , w.j . and delbeke , l. ( 2007 ) multidimensional unfolding by nonmetric multidimensional scaling of spearman distances in the extended permutation polytope .
_ multivariate behavioral research _ , 42(1 ) , 103132 .
_ proof _ : using the same notation as in choulakian ( 2014 ) , we designate@xmath527first , by ( 12 ) we have @xmath528from which we get,@xmath529second , by triangle inequality of the l@xmath93 norm we have@xmath530 \c ) the vector containing the first and only factor scores of the columns is @xmath536 where @xmath537 for @xmath352 in particular , we see that @xmath538 @xmath539 and @xmath379 for @xmath280 are equispaced .
so for an odd number of items @xmath380 for @xmath540 , we have @xmath541 _ proof _ : a ) a _ faithfully homogenous group _ consists of one response pattern , and its borda score values for a complete linear order of @xmath0 items , without loss of generality by relabeling of the items , will be @xmath542 .
then the nega coded correspondence table will have only two rows and @xmath0 columns@xmath364and @xmath543 will be of rank 1 ; that is , there is only one principle factor , for which we note the taxicab dispersion measure by @xmath367 for a fixed finite integer value of @xmath366by ( 18 ) the elements of @xmath318 are@xmath544and@xmath545so@xmath546the value of @xmath367 by theorem 1 b is@xmath547 \a ) necessary condition .
we have @xmath567 , \end{aligned}\]]where @xmath568 and @xmath569 for @xmath340 given that @xmath570 is a sum of @xmath571 positive terms , it is easy to see that * u*@xmath572 is the first tca principal axis , where @xmath573 is the characteristic function of @xmath574 that is , it has the value of 1 if @xmath575 and 0 otherwise .
so for @xmath368 and @xmath383 the first tca axis divides the set of items @xmath28 into 2 blocks such that @xmath384 with @xmath576 and @xmath577 that is , the partition of the set of items @xmath28 is _ faithful . _
furthermore , _ _ the borda scores of all voters are intra block permutations by definition of @xmath578 and @xmath579 .
\b ) sufficient condition .
we suppose that the partition of @xmath580 is faithful and there are no crossings between the 2 blocks @xmath581 and @xmath582 this implies that for @xmath583 @xmath584 for @xmath585 and @xmath586 for @xmath587 thus we get @xmath588 as in the proof of the necessary condition . | the traditional methods of finding mixture components of rank data are mostly based on distance and latent class models ; these models may exhibit the phenomenon of masking of groups of small sizes ; probably due to the spherical nature of rank data .
our approach diverges from the traditional methods ; it is directional and uses a logical principle , the law of contradiction .
we discuss the concept of a mixture for rank data essentially in terms of the notion of global homogeneity of its group components .
local heterogeneities may appear once the group components of the mixture have been discovered .
this is done via the exploratory analysis of rank data by taxicab correspondence analysis with the nega coding : if the first factor is an affine function of the borda count , then we say that the rank data are globally homogenous , and local heterogeneities may appear on the consequent factors ; otherwise , the rank data either are globally homogenous with outliers , or a mixture of globally homogenous groups . also we introduce a new coefficient of global homogeneity , ghc .
ghc is based on the first taxicab dispersion measure : it takes values between 0 and 100% , so it is easily interpretable .
ghc measures the extent of crossing of scores of voters between two or three blocks seriation of the items where the borda count statistic provides consensus ordering of the items on the first axis .
examples are provided .
key words : preferences ; rankings ; borda count ; global homogeneity coefficient ; nega coding ; law of contradiction ; mixture ; outliers ; taxicab correspondence analysis ; masking . |
predicting future developments of financial asset prices is an extremely difficult task .
certain problems , such as predicting the direction of movements , are nearly unsolvable . meanwhile , certain aspects can be easily predicted .
for instance , one can be sure that the prices will always fluctuate intermittently , largely due to people who believe they have found the winning algorithm , the philosopher s stone .
the analysis of the historic charts of price dynamics is referred to as technical analysis @xcite .
the essence of the technical analysis is the hypothesis that the patterns of historic data can forecast the future price movements . on the other hand , the efficient market hypothesis ( c.f .
@xcite ) states that security prices reflect fully all available information . in its century - long history
dated back to the work of bachelier @xcite , the financial analysis has made use of the both approaches .
the random walk hypothesis , standard deviation and the correlations between securities returns are the cornerstones of seminal papers in financial analysis @xcite .
econophysics has introduced various more elaborate , mostly non - linear tools of analysis ( c.f .
@xcite and references therein ) . here
we extend a recently developed technique of length - distribution of low - variability periods @xcite .
strongly non - linear systems , such as turbulent fluids and plasmas , granular media , biological and economical systems , etc .
, are typically characterized by scale - invariance and scaling laws .
so , it should be not surprising that such seemingly different disciplines like turbulence studies , biological physics , and econophysics have many common tools of data analysis .
for example , power - law distributions were first used in economics @xcite in the end of 19th century , and later found in a wide variety of systems ( often referred to as the zipf s law ) , c.f .
similarly , diverse systems are known to generate signals , which are self - affine , and hence , can be characterized by the hurst exponent , c.f .
further , the stable lvy distributions ( c.f .
@xcite ) have been found to be relevant to all the mention disciplines ; about truncated levy flights in econophysics , c.f . @xcite .
finally , multi - fractal formalism is by far the most popular tool for scale - invariant analysis of intermittent time - series , c.f .
. it should be noted , however , that in many cases ( including econophysical applications ) , there is no profound understanding of the origin of multi - fractality , and hence , the multi - fractal analysis is not necessarily the optimal method . indeed , multi - fractal formalism has been devised in the context of turbulence , and is specifically suited for systems with random multiplicative cascades @xcite .
meanwhile , in the case of such time - series as stock prices or heart rate variability signal , the presence of multiplicative cascades is not evident . therefore , _ there is a clear need for a deeper understanding of the character of intermittency in the case of financial time - series_. this problem can be approached by studying new independent and/or more general methods of scale - invariant analysis .
for instance , in the case of heart rate variability , it has been found that the whole time - period can be clusterised into self - similarly distributed segments of approximately constant heart rate @xcite . in order to address this problem ,
a new method has been suggested recently @xcite which is based on the analysis of the distribution of low - variability periods .
the low - variability period is defined as a time period with maximal length where consecutive relative changes in realizations of the time series are less than given threshold @xmath0 .
note that in addition to the financial assets , the low - variability period analysis has been applied to biological systems @xcite .
it has been shown @xcite that in the case of the multi - affine time series , the cumulative distribution function of the low - variability periods is in the form of a ( multi - scaling ) power - law .
since the opposite is not necessarily true , the low - variability period analysis is , indeed , a more universal method than the multi - affine analysis .
even if the time - series is actually multi - affine , the low - variability period analysis can be still useful , because _ ( a ) _ the power - law exponent is related to the multi - fractal dimension ; this circumstance provides an easy method for checking the assumption of multi - affinity ; _ ( b ) _ low - variability periods provide higher time - resolution of time series analysis @xcite .
this paper serves three main purposes .
_ first _ , we are going to apply the method of the analysis of low - variability periods to the data of trading volumes .
this analysis is motivated as follows .
similarly to the asset prices , the trading volumes are known to fluctuate intermittently , c.f .
@xcite . according to the mandelbrot s model of stock prices as a fractional brownian motion in multi - affine trading time @xcite
, one could expect that the time series of trading volumes are multi - affine .
the analysis of the length - distribution of low - variability periods of trading volumes serves as a test of this model .
_ second _ , we discuss the consequences of the presence of power laws of low - variability periods .
_ third _ , an attempt is made to generalize the method to multivariate time - series ( e.g. a stock price together with the trading volume ) .
it should be noted that in the case of multi - affine analysis , there is no simply interpretable way for such a generalization .
we start with a brief description of the method devised in ref .
@xcite , a low - variability period is defined as such a contiguous time - interval @xmath1 $ ] , which satisfies the conditions @xmath2 for @xmath3 ; it is assumed that the data sampling interval serves as the time unit .
the average of the trading volumes @xmath4 is taken over a time window of length @xmath5 : @xmath6 therefore , we have two control parameters : _ ( i )
_ @xmath5 the length of the averaging window , and _
( ii ) _ @xmath0 the variability threshold .
further , the cumulative length - distribution function of low - variability periods @xmath7 is introduced : @xmath7 is the number of the low - variability periods of length @xmath8 .
we speak about multi - scaling behaviour , if the following power - law is observed : @xmath9 where @xmath10 is a scaling exponent and @xmath11 is a constant . it has been shown @xcite that in the case of multi - affine time - series , the low - variability periods do follow a multi - scaling distribution law : @xmath12 where @xmath13 denotes the hlder multifractal spectrum of the local hurst exponents @xmath14 .
meanwhile , the presence of a scaling law ( [ power ] ) does not necessarily imply multifractality .
indeed , note that the hlder exponent @xmath13 can not exceed the topological dimensionality ( one ) , therefore the values @xmath15 are not related to multifractality .
it should be stressed that unequality @xmath16 does neither imply the lack of multifractality : in the case of multi - fractal time - series , @xmath16 can ( in principle ) be observed for @xmath17 , where @xmath18 is the dominant local hurst exponent , @xmath19 . for this range of parameters
, the unequality @xmath16 must be satisfied ( assuming that @xmath20 is a monotonous function of @xmath0 ) .
finally , if there is no data collapse @xmath21 for @xmath22 , the underlying time - series is certainly non - multifractal . in the case of currency rates , a reasonable data collapse [ according to eq .
( [ alphalog ] ) ] has been observed . in this section ,
the daily trading volume data of various stock indices are tested for multi - scaling and multi - fractality .
the stock market volumes are measured in the amount of shares traded on the exchange .
the data used in the analysis represent the daily closing prices and trading volumes and it is described in detail in the table [ data ] . .the data used in volume analysis [ cols="<,<,^,>",options="header " , ] as seen from fig .
[ histo ] and table [ alphas ] , the low - variability periods of the daily volume time series follow reasonably well a multi - scaling behaviour ( similarly to what has been observed in the case of stock prices @xcite ) . in fig .
[ logtau ] , the scaling exponent @xmath10 is plotted against the @xmath23 for dax and spx time series with @xmath24 days .
we can see that there is a departure from the multi - affine expectation there is no data collapse at range @xmath25 [ which would have been corresponding to eq .
( [ alphalog ] ) ] .
this observation is interpreted as follows . here
, each data point describes the amount of shares traded in respective stock exchange in a calendar day . the condition @xmath26 is satisfied only for very high thresholds ( @xmath27 ) .
typically , there are very few periods , when trading volumes would fluctuate with such a large amplitude .
therefore , the number of valid data points is very low . as a consequence ,
the scaling range is narrow , and the formally calculated scaling exponent is non - stationary ( depends on the particular realisation of the time - series ) .
this behaviour is different from what has been observed for currency time series ( when the results @xmath28 were more stationary , with a reasonable data collapse in @xmath29-plot ) .
so , we can conclude that while the trading volume data are intermittent and fluctuations are scale - invariant [ described by eq .
( [ power ] ) ] , the degree of intermittency is lower than in the case of currency fluctuations : the scaling exponent values tend to be always larger than one .
respectively , the multifractal pattern [ like described by eq .
( [ alphalog ] ) ] is not observable
. plotted against @xmath23 for spx and dax time series for @xmath24 days , title="fig:",width=491 ] [ logtau ] in the case of financial time - series , a crucial question is how to make forecasts .
the most useful kind of forecasts would give predictions of the direction of future prices .
however , these forecasts will be always very unreliable , the efficient market hypothesis denies such a possibility entirely ( c.f .
second law of thermodynamics ) .
meanwhile , the risk - related forecasts are completely possible ( and may prove to be useful ) .
therefore , a natural question arises : what are the risk - prediction - related consequences of the presence of the power - law distribution of low - variability periods ? more specifically ,
suppose we have had a low - variability period , which had lasted @xmath30 days ( and still goes on ) .
can we say something about the possibility of today being the last day of this `` silent '' period ?
in other words , what is the probability @xmath31 that the tomorrow s movement exceeds our pre - fixed threshold @xmath0 ? apparently , this probability is given by the ratio of _ ( a ) _
the number of those low - variability periods , the length of which is exactly @xmath30 , @xmath32 , and _
( b ) _ the number of those low - variability periods , which have length @xmath33 , @xmath34 . if @xmath30 is large , the difference @xmath35 can be calculated approximately as @xmath36 . upon applying eq .
( [ power ] ) we arrive at @xmath37 and @xmath38 . bearing in mind that @xmath39 , the final result is written as @xmath40 therefore , we have shown that the very presence of a power law of the low - variability periods has an interesting consequence : the probability that the tomorrow s price movement will be larger than the movements of @xmath30 preceding days , is inversely proportional to @xmath30 .
the predicted power law exponent is independent of the scaling exponent @xmath10 , i.e. we are dealing with super - universality .
this super - universality appears to be related to the super - universality of the scaling of direct avalanches in self - organised critical systems @xcite .
it allows us to make every day a series of forecasts : by scanning the values of the threshold parameter @xmath0 , we find the length @xmath41 of the current ( still ongoing ) low - variability period .
then , the probability that the tomorrow s movement @xmath42 remains below the threshold @xmath0 is estimated as @xmath43 .
note that the prefactor @xmath44 can be dropped , but will improve the forecast ; @xmath45 days means that only the prices of two subsequent days are compared .
in the technical analysis , various forms of multi - factor models have been used for some time @xcite . in particular
, it has been conjectured that in the case of trading volumes and price fluctuations , the higher - than average trading volumes generally `` confirm '' the price trend .
however , according to our best knowledge , scale - invariant methods have not yet been applied to the multi - factor studies .
the low - variability period analysis provides simply interpretable way of generalisation of multi - signal time series analysis . with the conventional multi - affine analysis ,
the interpretation of the results is significantly harder .
indeed , consider the wavelet transform @xmath46 , where @xmath47 denotes a coordinate and @xmath48 stands for the wavelet width .
for a multi - affine signal , the following scaling law is expected : @xmath49 . for a bi - variate analysis of two signals ,
this equation can be generalized as @xmath50 . here ,
@xmath51 and @xmath52 denote the wavelet transform of the two signals . for two dependent signals , @xmath53
; however , there is no clear way of interpreting the features of the scaling exponent @xmath54 .
therefore , in this section , the analysis of low - variability periods is generalised and the multi - factor model is proposed .
in particular , we address the above mentioned statement about the relationship between trading volumes and stock prices . in our previous calculations ,
the low - variability periods are defined either via the condition ( [ delta ] ) , or via the same unequality , with trading volume @xmath4 being substituted by stock price ( or index value ) @xmath55 @xcite : @xmath56 ) as the usage of the _ single - factor price model _ or _ method 0 _ ( for brevity ) .
similarly , the condition ( [ delta ] ) corresponds to the _ single - factor volume model _ or _ method 0 ( volume)_. there are two ways to generalize the concept of low - variability into multi - factor model : 1 . low - variability persists if both condition ( [ delta ] ) and ( [ m0 ] ) are satisfied ; 2 . low - variability persists if any of the conditions ( [ delta ] ) and ( [ m0 ] ) are satisfied .
it is clear that the number of low - variability periods is larger by using the latter option .
in fact , using the first option led us often to a very small number of low - variability periods ( this , of course , is related to the limited length of the time - series ) .
therefore , here we present only the results corresponding to the second definition . note that for multi - factor models with more than two inputs , the conditions similar to eq .
( [ m0 ] ) can be combined with any set of logical operators `` and '' and `` or '' . in what follows we use several definitions of the two - factor low - variability periods ( the application of these definitions
will be referred to as _ method 1 method 3 _ , respectively ) : @xmath57 @xmath58 @xmath59 equation ( [ m1 ] ) is symmetric with respect to the price and volume conditions .
the low - variability period is terminated as soon as the relative price change exceeds ( by modulus ) @xmath60 , and the relative volume change exceeds ( by modulus ) @xmath61 .
if the condition ( [ m1 ] ) is applied , the following asymptotic laws are expected : @xmath62 and @xmath63 such a behaviour is , indeed , observed for the time series of dax and spx .
the data used in the calculations is described in table [ data ] . in figures [ conv - m1dax ] and [ conv - m1spx ]
the scaling exponents @xmath64 are displayed for dax and spx time series by using @xmath65 days .
> from fig .
[ conv - m1dax ] , it is seen that for low values of @xmath66 , the low - variability condition for volume is almost always violated .
therefore , the scaling exponent is defined almost solely with the price parameter ( the curves with @xmath67 and @xmath68 are very close to each other ) . on the other hand , for high values of @xmath66 ,
the volume condition is typically satisfied , and the horizontal lines ( in @xmath20-@xmath60 space ) denote a low dependence on the price parameter .
these arguments apply to the dependence of the exponent @xmath64 on the price threshold parameters , as well , see fig .
[ conv - m1spx ] . .
_ method 0_. the scaling exponents @xmath64 are calculated for dax time series by using conditions ( [ m0 ] ) and ( [ m1 ] ) with @xmath65 days and @xmath69$],title="fig:",width=491 ] [ conv - m1dax ] .
_ method 0 ( volume)_. the scaling exponents @xmath64 are calculated for spx time series by using conditions ( [ delta ] ) and ( [ m1 ] ) with @xmath65 days and @xmath70$],title="fig:",width=491 ] [ conv - m1spx ] according to eq . ( [ m2 ] ) , the low - variability period is terminated when price change ( rise or drop ) is significant ( i.e. larger than the threshold parameter ) , and the volume increases faster than the threshold . equation ( [ m3 ] ) represents a definition , opposite to eq .
( [ m2 ] ) : the low - variability period is terminated when the price change exceeds the threshold parameter , and the volume decreases faster than the threshold .
these definitions are useful for studying the asymmetry between the volume rise and drop : if the multi - scaling exponent @xmath64 turns out to be different for methods 2 and 3 , there must be an asymmetry between those volume spike and squeeze events , which are accompanied by a large price variability .
indeed , the price condition in eqns ( [ m2 ] ) and ( [ m3 ] ) is the same ; so , the differences in the scaling exponent @xmath64 must be due to the different effect of the volume condition ( [ m2 ] ) and ( [ m3 ] ) .
let us refer back to method 1 [ eq .
( [ m1 ] ) ] . with this method
, the events terminating the low - variability periods represent a superposition of the respective events for the methods 2 and 3 [ eqns ( [ m2 ] ) and ( [ m3 ] ) ] .
so , if the scaling exponents calculated according to method 2 are very similar to the ones calculated according to method 1 , we can conclude , that the amount of the low - variability periods defined by method 3 is insignificant .
the same holds true if the scaling exponents of method 3 tend to be similar to the ones of method 1 .
the multi - factor scaling exponents @xmath64 are found for the dax data with @xmath65 days . in fig .
[ 2f - logtau ] , the scaling exponents @xmath71 of dax index are plotted against @xmath72 with _ ( a ) _ @xmath73 , _ ( b ) _
@xmath74 , _ ( c ) _ @xmath75 and _ ( d ) _ @xmath76 .
the methods 0 - 3 [ eqns ( [ m0])([m3 ] ) ] are used for definition of low - variability periods .
of dax time series are plotted against @xmath72 with _
( a ) _ @xmath73 , _ ( b ) _
@xmath74 , _ ( c ) _ @xmath75 and _ ( d ) _ @xmath76,title="fig:",width=529 ] [ 2f - logtau ] by giving a small value to the thresholds @xmath66 and
@xmath60 in any of eqns ( [ m0])([m3 ] ) , the number of longer low - variability periods becomes small .
therefore , we have a large number of short low - variability periods ( compared to high @xmath0 values ) ; this leads to high values of the scaling exponent . in fig .
[ 2f - logtau]a , the threshold value of @xmath66 is set to very low level of @xmath77 . from eq .
( [ m1 ] ) it can be seen that then the two - factor model leads to results which are very similar to the single - factor ( i.e. price factor ) model .
likewise , the larger the volume threshold @xmath66 , the larger the difference between the @xmath20@xmath60-curves of the single- and multi - factor models ( at @xmath76 , the curves are rather dissimilar ) . an important issue is the difference between the curves of methods 2 and 3 .
one can notice that there is a different behaviour at the small values of the parameter @xmath60 , an evidence of the asymmetry between volume rise and drop .
one can also notice that the scaling exponents calculated according to the method 3 [ eq .
( [ m3 ] ) ] are lower than the ones calculated according to methods 1 and 2 [ eqns ( [ m1 ] ) and ( [ m2 ] ) ] .
meanwhile , the difference between the outcomes of the methods 1 and 2 [ eqns ( [ m1 ] ) and ( [ m2 ] ) ] is minor .
therefore , we conclude , that high price variability is typically accompanied by increasing volume .
this conclusion is independent of the price / volume pre - history ( i.e. is valid both for short and long low - variability periods ) . in this paper
, we have not analysed the problem of higher - rank multi - factor models
. however , this can be useful for e.g. multi - stock data analysis , where each stock price provides an independent input stream .
this situation will be addressed in further studies .
the concept of low - variability periods has been proven to be useful for various econophysical issues ( not just limited to the scope of stock prices / indices and currency exchange rates ) .
so , we found that the time series of stock trading volumes obey multi - scaling properties , similarly like the price data .
however , while the multi - scaling exponent of the price time series follows a pattern , characteristic to the multi - affine data , in the case of trading volumes , there is a clear departure from that pattern ( one can say that the fluctuations are less intermittent ) .
further , we have shown that the presence of the multi - scaling distribution of the length - distribution of the low - variability periods gives rise to a super - universal scaling law for the probability of observing next day a large price movement .
this probability is inversely proportional to the length of the ongoing low - variability period , a fact which can be used for risk forecasts .
finally , the multi - factor model is proposed for time series analysis . in this paper , only the simplest two - factor model is described and applied to stock price and volume data .
the low - variability periods of multi - variate time series can be defined in different ways ; for instance , the threshold conditions applied to the single data streams can be combined by logical `` and '' , as well as by logical `` or '' . in the our case of price and volume data , three different definitions of low - variability periods
have been applied ( in order to study the asymmetry between the volume rise and drop , we have also applied sign - dependant threshold conditions ) .
this analysis led us to the conclusion that high price variability is typically accompanied by increasing trade volume , independently of the prior events of the market . in the light of this observation , the common thesis of technical analysis
, `` increased trading volumes confirms the price trend '' , becomes less useful .
indeed , most of the significant price jumps are accompanied by increased trading volumes ; so , almost all the `` price trends '' pretend to be `` confirmed '' .
the support of estonian sf grant no .
5036 is acknowledged .
we would also like to thank prof .
jri engelbrecht for fruitful discussions .
e. fama , journal of business 38 ( 1965 ) 34 e. fama , journal of finance 25 ( 1970 ) 383 e. fama , journal of finance 46 ( 1991 ) 1575 l. bachelier , theorie de la speculation .
( doctoral dissertation in mathematical sciences , faculte des sciences de paris , defended march 19 , 1900 h. markowitz , journal of finance 7 ( 1952 ) 77 w. sharpe , journal of finance 19 ( 1964 ) 425 f. black , m. scholes , journal of political economy 81 ( 1973 ) 637 r.n .
mantegna , h.e .
stanley , an introduction to econophysics : correlations and complexity in finance , cambridge university press , cambridge , 1999 m. ausloos , ph .
bronlet , physica a 324 ( 2003 ) 30 x. gabaix , p. gopikrishnan , v. plerou and h.e .
stanley , nature 423 ( 2003 ) 267 xavier gabaix , power laws and the origins of the business cycle , mit , department of economics and nber , august 20 , 2003 y. fujiwara , c. di guilmi , h. aoyama , m. gallegati , w. souma , physica a 335 ( 2004 ) 197 y. fujiwara , physica a 337 ( 2004 ) 219 b.b .
mandelbrot , the fractal geometry of nature , w.h.freeman , new york , 1982 n. vandewalle , m. ausloos , eur .
j. b 4 ( 1998 ) 257 r. benzi , g. paladin and a. vulpiani and m. vergassola , j. phys . a : math 17 ( 1984 ) 3521 p. bernaola - galv ' an , p.ch .
ivanov , l.a.n .
amaral and h.e .
stanley , phys .
87 ( 2001 ) 168105 j. kalda , m. skki , m. vainu , m. laan , arxiv : physics/0110075 v1 26 oct 2001 , m.skki , j. kalda , m. vainu , m. laan , physica a 338 ( 2004 ) 255 m. bachmann , j. kalda , j. lass , v. tuulik , m. skki , h. hinrikus , method of nonlinear analysis of electroencephalogram for detection of the effect of low - level electromagnetic field , accepted ( 2004 ) by : med .
comput .
p. gopikrishnan , v. plerou , x. gabaix , h.e .
stanley , phys . rev .
e 62 ( 2000 ) 4493 m. ausloos , k. ivanova , in : h. takayasu ( ed . ) , the applications of econophysics , proceedings of the second nikkei econophysics symposium , springer verlag , berlin , 2004 , p. 117
s. maslov , phys .
74 ( 1995 ) 562 | the scaling properties of the time series of asset prices and trading volumes of stock markets are analysed .
it is shown that similarly to the asset prices , the trading volume data obey multi - scaling length - distribution of low - variability periods . in the case of asset prices , such scaling behaviour can be used for risk forecasts : the probability of observing next day a large price movement is ( super - universally ) inversely proportional to the length of the ongoing low - variability period . finally , a method is devised for a multi - factor scaling analysis .
we apply the simplest , two - factor model to equity index and trading volume time series . , econophysics , multi - scaling , low - variability periods 89.65.gh , 89.75.da , 05.40.fb , 05.45.tp , |
quantum phase transitions ( qpts ) of a quantum many - body system have been attracting the persistent interest of physical researchers in recent years . due to the diversity of quantum phases and qpts
, finding universal ways or methods to characterize qpts is very meaningful and urgent . from the viewpoint of landau - ginzburg theory which has been widely accepted and known in condensed matter physics @xcite ,
qpt is connected with the corresponding order parameter and symmetry breaking . however , there are also some qpts which can not be well understood under the landau - ginzburg paradigm , such as the topological phase transitions @xcite and beresinskii - kosterlitz - thouless ( bkt ) phase transitions beresinskii , kosterlitz .
recently , an increasing research effort has been focused on the role of ground - state fidelity in characterizing qptsgu_review , htquan , zanardi06 , hqzhou , youwl07,zanardi07prl , schen07pre , schen08pra , buonsante , mfyang , zhouprl . as a basic concept in quantum information science ,
the fidelity measures the similarity between two states and is simply defined as modulus of their overlap @xcite .
the fidelity approach provides us a novel way to understand qpts from the viewpoint of quantum information theory .
so far qpts in various quantum many - body systems @xcite have been shown to be well characterized by the ground - state fidelity or fidelity susceptibility which is the leading term of the fidelity youwl07,zanardi07prl .
generally , one may expect that the structure of the ground states at the different phases is basically different and should reveal itself by some sort of singular behavior in the ground state fidelity or the fidelity susceptibility at the transition point @xcite . despite its great successes of application in various systems , this intuitive idea turns out to be not complete @xcite .
although the fidelity and the fidelity susceptibility can be used to describe first- and second - order qpts@xcite , as well as the topological qpts @xcite successfully , nevertheless there are also some ambiguous cases for that both the two methods mentioned above do not work very effectively @xcite . very recently ,
the controversial issue of bkt phase transition and ground state fidelity has been studied in ref .
@xcite from a perspective of matrix product states which essentially depend on a classical simulations of quantum lattice systems @xcite . in case
that the leading term of the fidelity ( fidelity susceptibility ) works not very effectively , the higher order term in the fidelity may be worth studying .
up to now , there is still lack of literature concerning this part of the fidelity . here
, in this paper , we make an attempt on investigating the effect of higher order term of the fidelity on the characterization of the bkt - type phase transition happened in the heisenberg chain with next - nearest - neighbor ( nnn ) interaction @xcite .
we will show that although the fidelity and fidelity susceptibility can not effectively characterize the bkt - type phase transition point for the heisenberg chain with nnn interaction , the higher order term of the fidelity gives a good attempt on detecting such a transition .
our paper is organized as follows . in sec .
ii , we display the formulism of the higher order term of the fidelity .
the subsequent section is devoted to the calculation of the higher order term of the fidelity for the model of heisenberg chain with nnn interaction and show its connection to the quantum phase transition of the system .
a brief summary is given in sec .
sec : sum .
as usual , the ground state fidelity is defined as the modulus of the overlap between @xmath0 and @xmath1 , i.e. @xmath2 where @xmath3 is the ground - state wavefunction of hamiltonian @xmath4 , @xmath5 is the driving parameter and @xmath6 is a small deviation in the parameter space of @xmath5 .
the fidelity susceptibility denotes only the leading term of the fidelity .
straightforwardly , one can get the higher order term of the fidelity following similar expansion in deriving the fidelity susceptibility youwl07 . by using the taylor expansion ,
the overlap between two wavefunction @xmath7 and @xmath8 can be expanded to an arbitrary order of @xmath9 , i.e. @xmath10 therefore , the fidelity becomes @xmath11 we note that @xmath12 and use the relation for a given @xmath13 @xmath14 then we can simplify the expression of ( [ eqf2 ] ) into @xmath15where @xmath16 with the projection operator @xmath17 defined as @xmath18 .
it is easy to check that @xmath19 is zero and @xmath20 the fidelity susceptibility @xcite .
next we shall consider the third order fidelity @xmath21 and apply it to judge the phase transition in the spin chain model with nnn exchanges .
alternatively , one can directly derive the expression of @xmath21 from the perturbation expansion of the gs wavefunction . according the perturbation theory , the gs wavefunction , up to the second order , is @xmath22the 3rd order term @xmath21 , which is proportional to the 3rd order derivative of gs fidelity , can be then directly extracted from eq .
( [ f2simple ] ) : @xmath23 eqs .
( [ eq : higherorderdiff ] ) and ( [ eq : higheroderperturb ] ) present the main formulism of the higher order expansion of the fidelity .
so far the explicit physical meaning of the high order term in the fidelity is still not clear .
the expression of 3rd fidelity bears the similarity to its correspondence of the 3rd derivative of gs energy which has the following form @xmath24 obviously , the 3rd fidelity is more divergent than the 3rd derivative of gs energy .
similar connection between the fidelity susceptibility and 2nd derivative of gs energy has been unveiled @xcite .
generally the @xmath25-th order fidelity is much more divergent than its counterpart of @xmath13-th order derivative of gs energy , therefore an @xmath13-th order qpt can be certainly detected by the @xmath13-th order fidelity .
however , this conclusion does not exclude the possibility that @xmath13-th order fidelity can detect a even higher order or infinite order qpt .
a concrete example has been given in ref .
mfyang08 , where a qpt of higher than second order was singled out unambiguously by using the fidelity susceptibility despite the corresponding second derivative of the ground - state energy density showing no signal of divergence .
so far no example of bkt - type qpt unambiguously detected by fidelity susceptibility has been given .
next we shall attempt to apply the third - order fidelity to study the bkt - type transition in a the spin chain model with nnn exchanges .
now we turn to the one - dimensional heisenberg chain with the nnn coupling described by the hamiltonian @xmath26where @xmath27 denotes the spin-1/2 operator at the @xmath28th site , @xmath29 denotes the total number of sites .
the driving parameter @xmath30 represents the ratio between the nnn coupling and the nearest - neighbor ( nn ) coupling .
the gs properties of the model ( [ ham ] ) has been widely studied by both analytical method @xcite and numerical method okamoto , castilla , rchitra , srwhite96 .
the qpt driven by @xmath30 is well understood .
the driving term due to @xmath30 is irrelevant when @xmath31 , and the system flows to a spin fluid or luttinger liquid with massless spinon excitations .
as @xmath32 , the frustration term is relevant and the ground state flows to the dimerized phase with a spin gap open @xcite .
the transition from spin fluid to dimerized phase is known to be of bkt type @xcite , for which the transition point was hard to be determined numerically due to the problem of logarithmic correction @xcite .
the critical value of @xmath33 has been accurately determined by various numerical methods @xcite .
the gs fidelity for the model ( [ ham ] ) has been studied in ref .
schen07pre and also in ref .
@xcite in terms of operator fidelity .
no singularities in the gs fidelity or operator fidelity around @xmath34 have been detected for the system with different sizes , which implies that the gs fidelity may be not an effective characterization of the bkt - type qpt in this model .
the bkt - type qpt is a infinite order phase transition where the @xmath13-th order derivative of gs energy is continuous . , @xmath35 .
, width=340 ] in light of the higher - order fidelity being more powerful than its energy judgement , we study the possibility for detecting the infinite - order bkt - type qpt via the 3rd order fidelity and focus on the qpt in the spin chain with nnn interactions as a concrete example .
we first calculate the gs wave functions by using the numerical exact diagonalization method for finite size system , and thus the fidelity susceptibility and the 3rd order fidelity can be extracted from the overlap of neighboring gs wave functions . in fig.1 , we display the fidelity susceptibility for systems with different sizes .
we observe that no an obvious peak for the fidelity susceptibility is detected in a wide range of the parameter @xmath36 .
this result suggests that the transition point for the bkt - type qpts can not be very effectively characterized by the fidelity susceptibility either for a finite - size system .
the bkt - type phase transition generally is an infinite order phase transition for which the infinite order derivatives of the ground - state energy is continuous . a good example with exact proof is the bkt - type transition happened in the antiferromagnetic xxz spin chain model yangcn . in the bkt - type transition point
, it has been proven analytically that all the @xmath13-th order derivatives of ground state energy is continuous @xcite . since the @xmath13-th order fidelity is much divergent than its correspondence of derivative of the ground - state energy
, one might expect that there exists the possibility that the @xmath13-th order fidelity is divergent even its @xmath13-th order energy derivative is continuous . to see whether a higher order fidelity works better than fidelity susceptibility in detecting the bkt - type qpt happened in this model , we calculated the 3rd order fidelity versus the driving parameter as shown in fig .
2 . it is clear that a peak is developed in the 3rd order fidelity and the location of peaks tends to get close to the side of transition point @xmath37 with the increase of lattice size . to extrapolate the @xmath38 in the infinite size limit
, we analyze the finite size scaling of position of peak in the fig .
when the system size comes to infinity , the extrapolated value of the phase transition point is @xmath39 , which , within the scope of fitting error , agrees well with @xmath40 obtained by highly accurate numerical methods okamoto , castilla , rchitra , srwhite96 .
we have shown the formulism for the high order of the fidelity in detail and applied it to a concrete model , _
i.e. _ , the one dimensional heisenberg chain with nnn interaction .
we first calculate the ground - state wavefunction of the system by exact diagnolization method , and then extract fidelity susceptibility and the third order of the gs fidelity .
we find that despite the gs fidelity and the fidelity susceptibility being not a very effective detector , the bkt - type phase transition happened in this spin chain model might be effectively detected by the 3rd order term of the gs fidelity for finite - size system .
although the physical meaning of the higher order term of the gs fidelity hasnt been deeply understood , we wish that our observation would stimulate further studies on this issue .
this work is supported by nsf of china under grant no .
10821403 , programs of chinese academy of sciences , national program for basic research of most , china and the earmarked grant research from the research grants council of hksar , china ( project no .
cuhk 400807 ) . | in this article , we study the high order term of the fidelity of the heisenberg chain with next - nearest - neighbor interaction and analyze its connection with quantum phase transition of beresinskii - kosterlitz - thouless type happened in the system .
we calculate the fidelity susceptibility of the system and find that although the phase transition point ca nt be well characterized by the fidelity susceptibility , it can be effectively picked out by the higher order of the ground - state fidelity for finite - size systems . |
blends of emission lines are often among the most prominent broad emission features in the ultraviolet ( uv ) and optical spectra of broad - lined active galactic nuclei ( agns ) , and the integrated flux of emission in quasars can be greater than that of any other single emission line , including ly @xmath6 @xcite . despite more than three decades of observational and theoretical effort , the physical conditions that give rise to emission have remained very difficult to determine .
this is due in part to the complex energy - level structure of the fe@xmath7 ion and the very large number of individual emission lines that appear in broad blends .
recent spectral synthesis models for the fe@xmath7 ion include hundreds of energy levels and tens of thousands of individual transitions @xcite .
a variety of processes in the broad - line region ( blr ) can contribute to line production , including collisional excitation as well as continuum and line fluorescence @xcite .
it is not yet fully determined whether the production in agns occurs in gas heated solely by photoionization , or whether collisional ionization might play a significant or even dominant role @xcite .
a particularly promising recent development has been the realization that anisotropy in the emission from high column density clouds can have a strong impact on the observed spectrum , and models for the emission from the shielded side of photoionized clouds have shown significant improvements in fitting observed spectra @xcite .
the emission lines are closely linked to several fundamental issues in agn physics and phenomenology . from principal - component analysis of quasar spectra ,
the relative strength of emission is one of the major characteristics of `` eigenvector 1 ''
@xcite , the component which accounts for much of the variance among the quasar population .
empirically , eigenvector 1 represents the anticorrelations between and [ ] equivalent widths , and between strength and h@xmath0 width .
the fundamental physical driver of the eigenvector 1 sequence is the eddington ratio @xmath8 @xcite , but the physical mechanism responsible for the increase in strength with eddington ratio is not fully understood .
it may be linked to the presence of a soft x - ray excess at high eddington ratio , which would produce a larger partially ionized zone of warm , -emitting gas ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , while orientation could play a secondary role @xcite .
@xcite argue against the x - ray excess as the primary driver of strength , proposing instead that differences in gas - phase iron abundance , driven by selective depletion of iron onto grains , are primarily responsible for the wide range of observed line strengths in agns . in another interpretation proposed by @xcite and @xcite , the competing forces of gravity and radiation pressure set a critical column density for clouds to remain gravitationally bound within the blr , and at higher eddington ratio
, the higher column density of surviving blr clouds would produce a larger /h@xmath0 flux ratio . for agns having extremely strong emission ,
it has also been suggested that the shock - heated gas associated with circumnuclear star formation may be responsible for the enhancement @xcite .
furthermore , emission is potentially valuable as a tracer of chemical evolution .
the strength of the uv emission relative to in quasar spectra has been used as a proxy for the iron to @xmath6-element abundance ratio in the blr ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , although the sensitivity of / line ratios to factors such as gas density and microturbulence means that observed line ratios are not straightforward indicators of the underlying abundance ratios @xcite .
improved understanding of the physical conditions responsible for emission could have important ramifications for elucidating the history of metal enrichment in the densest regions of the universe at high redshift .
a key quantity of interest is the spatial scale of the -emitting region .
however , there are very few measurements available that directly constrain its size .
@xcite carried out reverberation mapping of the uv lines in ngc 5548 and found a lag of about 10 days , similar to that of ly@xmath6 , indicating that the uv lines originate from within the blr .
this has been the only measurement of a reverberation lag for the uv lines to date . aside from reverberation mapping ,
one additional direct constraint exists : a spectroscopic microlensing study of a lensed quasar by @xcite found evidence that the emission originates largely in the outer portion of the blr .
previous attempts at reverberation mapping of the optical blends have generally not led to clear detections of reverberation lags . for the well - studied agn ngc 5548
, @xcite examined 13 years of monitoring data and found that the flux responded to continuum changes on timescales of less than several weeks , but the available data did not allow for a definite measurement of the lag .
@xcite measured light curves for the blends in ark 120 and found a long - term variability trend that followed the continuum changes , but the data did not yield a significant cross - correlation lag .
they concluded that the emission region in ark 120 was likely to be several times larger than the h@xmath0-emitting zone of the blr .
an alternative possibility was that the emission might be powered by collisional excitation rather than by photoionization , although they were unable to fit the emission blends well with collisional excitation models .
@xcite measured the light curve for pg 1700 + 518 using monitoring data from @xcite . while they were able to detect evidence of reverberation in the lines ,
the cross - correlation analysis did not yield a highly significant peak , and the measured lag of @xmath9 days was very uncertain .
variability of the optical emission has been examined in other agns ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but most datasets have not been suitable for measurement of reverberation lags .
the overall picture that emerges from these studies is that while optical emission does respond to continuum variations at least over long timescales , it tends to have a lower amplitude of variability than h@xmath0 , and it does not generally show a clear reverberation lag signature . in the absence of direct constraints on the relative sizes of the and h@xmath0 emission regions , indirect clues have come from line - profile measurements . @xcite and @xcite found that the velocity widths of the lines are generally similar to that of h@xmath0 , suggesting that and h@xmath0 originate from the same region within the blr .
subsequent work has uncovered subtle systematic differences between h@xmath0 and optical widths , which hint at an origin for in the outer portion of the blr or perhaps within an `` intermediate - line region '' corresponding to the transition between the blr and the dusty torus @xcite . for a sample of 4000 sloan digital sky survey ( sdss )
quasar spectra , @xcite found that the full - width at half maximum intensity ( fwhm ) of the optical lines is typically about @xmath10 fwhm(h@xmath0 ) , albeit with substantial scatter , implying a size for the -emitting region that is typically about twice as large as the h@xmath0-emitting zone of the blr . here , we present new reverberation - mapping results for two nearby seyfert galaxies , ngc 4593 and mrk 1511 , from the lick agn monitoring project 2011 .
section 2 gives a brief overview of our observing campaign .
section 3 describes the fitting method used to isolate the continuum and emission - line components in the spectroscopic data and the measurement of light curves . in section 4
we describe the cross - correlation lag measurements and the dependence of the lags on the choice of template , and section 5 presents a discussion of the results . in an appendix , we present a brief discussion of results based on additional newly - available templates . while the reverberation lag measurements are modestly sensitive to the choice of template , both of these objects clearly show evidence of reverberation in response to continuum variations .
initial results from the lick agn monitoring project 2011 were previously presented by @xcite and @xcite , and full details of the spectroscopic and photometric observing campaigns , analysis procedures , and light - curve data for all targets will be presented in a forthcoming series of papers . here , we briefly review the key aspects of the observing program . from early march through mid - june 2011 , we obtained queue - scheduled @xmath1-band images of ngc 4593 and mrk 1511 using the 0.76 m katzman automatic imaging telescope ( kait ) at lick observatory @xcite , the 0.9 m telescope at the brigham young west mountain observatory ( wmo ) , the faulkes telescope south at siding spring observatory , and the palomar 1.5 m telescope @xcite .
our goal was to obtain nightly imaging for each target , but weather and scheduling issues led to some gaps in coverage , particularly during the initial portion of the campaign .
the temporal sampling cadence of our observations is described in table [ lcstats ] .
exposure times were typically 180300 s. all images were bias - subtracted and flattened , and cosmic - ray hits were removed using the la - cosmic routine @xcite .
image - subtraction photometry was carried out using a version of the isis code @xcite modified by the high-@xmath11 supernova search team @xcite and with additional modifications by w. li , and also using the ` hotpants ` package by a. becker , which is based on the methods described by @xcite .
for each telescope , a high - quality template image was chosen , and the template was then aligned with each night s image and convolved with a spatially varying kernel to match the point - spread function of that image . after subtracting the scaled template image , the variable agn flux is left as a point source in the subtracted image , allowing for aperture photometry using the iraf daophot package . the photometric aperture radius used for each telescope
was set to match the average point - source fwhm for images from that telescope .
lccc ngc 4593 photometry & 75 & 0.95 & 1.12 + ngc 4593 photometry ( averaged ) & 57 & 1.01 & 1.48 + ngc 4593 spectroscopy & 43 & 1.02 & 1.81 + mrk 1511 photometry & 71 & 1.00 & 1.51 + mrk 1511 photometry ( averaged ) & 59 & 1.05 & 1.82 + mrk 1511 spectroscopy & 40 & 1.01 & 1.94 + light curves were initially constructed separately with the data from each telescope , and then combined together to assemble a single light curve for each agn .
normalization was done by modeling the variability using gaussian processes to obtain a finely - sampled model version of the light curve for each agn , following the same procedure described by @xcite .
additive and multiplicative scaling factors were applied to match the modeled light curve to each telescope light curve , using the wmo light curve as the reference data for the model .
the normalization code simultaneously constrained all scaling factors to obtain the best fit between all the telescope light curves using a markov chain monte carlo algorithm .
then , scaling factors determined from the modeling were applied to the image - subtraction light curves , to normalize the data from all telescopes to a common flux scale .
the final light curve was prepared by taking a weighted average of observations taken within 6 hours of one another .
the @xmath1-band light curves are shown in figure [ figlightcurves ] , in arbitrary units of differential flux .
our spring 2011 observing campaign consisted of 69 nights at the lick 3-m telescope between 2011 march 27 and june 13 ut . observations were done using the kast dual spectrograph @xcite .
a d55 dichroic was used to separate the blue and red beams with a crossover wavelength of about 5500 . in this paper , we discuss only measurements from the blue arm of the spectrograph , where we used a 600 lines mm@xmath12 grism over @xmath135520 at a scale of 1.0 pixel@xmath12 .
all observations of these two agns were done with a 4-wide slit oriented at a position angle of 45 .
standard calibration frames including arc lamps and dome flats were obtained each afternoon , and flux standards were observed during twilight .
exposure times were normally @xmath14 s and @xmath15 s for ngc 4593 and mrk 1511 , respectively . the weather at mt .
hamilton was somewhat worse than average during spring 2011 ; we were able to observe ngc 4593 on 43 of the 69 nights , and mrk 1511 on 40 nights .
spectroscopic reductions and calibrations followed standard methods implemented in iraf and idl .
a large extraction width of 103 was used in order to accommodate the full extent of the agn spatial profiles observed on nights with poor seeing .
error spectra were extracted and propagated through the full sequence of calibrations . in the reduced spectra , the median signal - to - noise ratio ( s / n ) per pixel in the range 46004700 is 122 for ngc 4593 and 71 for mrk 1511 .
the reduced spectra were first normalized to a uniform flux scale by employing the procedure of @xcite .
this method applies a flux scaling factor , a linear wavelength shift , and a gaussian convolution to each spectrum in order to minimize the residuals between the data and a reference spectrum constructed from several of the best - quality nights .
the scaling is determined using a wavelength range containing the [ ] @xmath16 line , which is assumed to have constant flux .
the upper panels of figure [ figmean - veron ] show the mean of all scaled spectra for each object . to assess the accuracy of the scaling procedure
, we measured the light curve of the [ ] @xmath16 line in the scaled spectra and calculated the residual `` excess '' scatter @xmath17 in the [ ] light curve .
we follow the usual definition for the normalized excess variance @xcite : @xmath18,\ ] ] where @xmath19 is the total number of observations , @xmath20 is the mean flux , and @xmath21 and @xmath22 are the individual flux values and their uncertainties
. then @xmath17 gives a measure of the fractional scatter in the light curve due to random errors in flux scaling , over and above the scatter that would be expected based on the propagated photon - counting uncertainties .
for ngc 4593 and mrk 1511 , we obtain an excess scatter of @xmath23 and 2.0% , respectively .
these low values indicate that the scaling procedure worked well . as we describe below
, these measures of residual scatter will be added to the error budget of the emission - line light curves to give a more realistic estimate of the total uncertainties .
traditionally , broad emission - line fluxes in reverberation - mapping data have been measured by choosing continuum regions on either side of an emission line , fitting a line to the continuum , and integrating the flux above that line ( e.g. , * ? ? ?
. however , in agn spectra there may be no pure continuum regions at all surrounding h@xmath0 , because , , and other emission features are present .
choosing continuum regions that contain some emission - line flux could potentially bias a reverberation lag measurement , particularly for velocity - resolved reverberation signals in the faint high - velocity wings of h@xmath0 .
an alternative approach is based on decomposition of spectra into continuum and emission - line components , so that the flux contributions of individual emission features can be isolated .
our multi - component fitting method was previously applied to mrk 50 @xcite and is similar to the method used by @xcite .
we carry out fits over the wavelength range extending from just blueward of the h@xmath24 line ( @xmath25 ) up to 5470 near the dichroic cutoff . over this wavelength range ,
each nightly spectrum was fit with a model consisting of several components : a power - law featureless continuum ( fc ) , a starlight template broadened in velocity by convolution with a gaussian , emission lines including h@xmath0 , [ ] @xmath26 , @xmath27 , and three lines ( 4471 , 4922 , and 5016 ) , and an template broadened in velocity by convolution with a gaussian .
for the starlight template , we use an 11 gyr - old simple stellar population model at solar metallicity from @xcite .
we experimented with adding younger stellar population components , but the 11 gyr model provided a sufficiently good fit that more complex stellar population models are not warranted , and the flux of a young population component is usually not very well constrained over this limited wavelength range .
we used two different templates to model the lines , from boroson & green ( 1992 ; hereinafter bg92 ) , and vron - cetty et al .
( 2004 ; hereinafter v04 ) .
the overall quality of the fits is very similar with the two templates , but there are some subtle differences that affect the reverberation lag measurements .
we find that the v04 template produces better results than the bg92 template ( see section [ sec : templates ] ) , but we describe measurements and reverberation lags derived from both sets of spectral fits in order to illustrate the dependence of the results on the choice of template . the broad component of h@xmath0 , as well as the [ ] lines , are represented by fourth - order gauss - hermite functions , while gaussians were used to represent weaker features including narrow h@xmath0 , the broad and narrow components of , and the lines .
we found that the complex shape of the h@xmath24 + [ ] @xmath28 blend was difficult to model , and instead of adding more model components we simply set the wavelength range subtended by this feature to have zero weight in the fit ( 42804395 rest wavelength ) .
additionally , a uniform foreground extinction was applied to the model spectrum using the @xcite reddening law .
this extinction correction represents the combined effects of galactic foreground extinction , extinction within the agn host galaxy , and wavelength - dependent slit losses resulting from the nonparallactic slit orientation @xcite . since these slit losses differ from night to night , we allowed the value of @xmath29 to vary freely in each fit .
the full model includes 29 free parameters , and was fitted to the observed spectrum by @xmath30 minimization using a levenberg - marquardt technique @xcite . for each galaxy ,
the fit was first carried out on the high - s / n mean spectrum , and then the parameters from the best fit to the mean spectrum were used as starting parameter estimates for the fit to each nightly spectrum .
for both objects , the fit optimization always drove the flux of @xmath31 to zero , using either template .
however , inclusion of the other lines at 4922 and 5016 does significantly improve the fit quality compared to model fits that omit these lines , at least when using the v04 template .
previous work has shown that these two lines can be important contributors to the `` red - shelf '' region redward of h@xmath0 @xcite .
the red shelf also contains some emission @xcite and the i zw 1 templates contain some flux in this region .
however , the relative amount differs between the bg92 and v04 templates , such that the v04 template has relatively weaker emission in the h@xmath0 red shelf , and this leads to noticeably different fitting results for the two templates .
using the v04 template , we found that the template alone did not produce an adequate fit to the red shelf unless the lines were added as separate components .
when fits are carried out using the bg92 template instead , the fluxes of the @xmath32 components in the h@xmath0 red shelf region go to zero or nearly zero , and essentially all of the flux in the h@xmath0 red shelf is taken up by the template .
another difficulty in fitting this region is that the @xmath33 line is degenerate with the red wing of h@xmath0 ; in order to achieve a consistent deblending of this spectral region into h@xmath0 and flux , we constrained the 4922 and 5016 lines to have identical fluxes ( following * ? ? ? * ) and identical velocity widths . in the fits
, these two components essentially track the same shape as two broad bumps in the broadened templates . due to the degeneracy between the several features contributing to the h@xmath0 red shelf , we are unable to determine a physically unique decomposition of this region into separate contributions from h@xmath0 , , , and [ ] .
the fits are able to match the observed spectra well in this region using either template , but the slight differences in the red shelf decompositions may be partly responsible for the dependence of the reverberation lags on the choice of template .
the strength and broadening of the component in the fit are primarily determined by the prominent blend around @xmath344700 containing several transitions from multiplets 37 and 38 .
the velocity - broadening width of was allowed to vary independently to optimize the fit to each nightly spectrum , but the variation from night to night was relatively small . from the best fit to each night
s data using the v04 template , the median velocity broadening ( i.e. , the dispersion of the gaussian broadening kernel ) applied to the template is @xmath35 km s@xmath12 for ngc 4593 and @xmath36 km
s@xmath12 for mrk 1511 .
the v04 template itself was constructed using lorentzian profiles for each line , with a width of fwhm = 1100 km s@xmath12 to match the line profiles in i zw 1 @xcite .
we find a similarly small night - to - night scatter in the width of the broadening kernel when using the bg92 template ( @xmath37 km s@xmath12 ) , although the overall amount of broadening is substantially smaller due to the different template structure .
the bg92 template is characterized by an intrinsic fwhm@xmath38900 km s@xmath12 ( e.g. , * ? ? ?
* ) , but as an empirically - derived template it contains features with a range of widths , and shapes that do not correspond precisely to simple gaussian or lorentzian profiles . as a result , it is not straightforward to compare the inferred widths as measured using the two different templates , but for each template the fitting procedure consistently converges on a value for the broadening that varies only by a few percent from night to night .
we comment further on the profile widths in section [ discussion ] .
figures [ figmean - veron ] and [ figmean - boroson ] show the mean spectrum for each agn with its best - fitting model , and the individual fit components , for the fits done using the v04 and bg92 templates .
the root - mean - square ( rms ) spectrum is constructed by taking the standard deviation of the scaled spectra ( e.g. , * ? ? ?
in each of these figures we show the rms spectrum , as well as a modified version constructed by subtracting the agn fc and stellar - continuum components from each nightly spectrum before calculating the rms .
removing the continuum components gives an improved rms spectrum by eliminating small residual stellar absorption - line features @xcite , and the continuum - subtracted rms spectrum more accurately depicts the rms variability profiles of the emission lines . fits done using the bg92 template assign relatively more flux to than fits with the v04 template .
for the bg92 fits , the increased flux ( at the expense of the fc component ) results in a higher rms flux in the continuum - subtracted rms spectra . light curves were measured for individual emission lines from continuum - subtracted spectra . for h@xmath0 and the h@xmath24 + [ ] blend
, we produced a residual spectrum by subtracting all other model components , and then measured the emission - line light curves by summation of the residual flux .
the h@xmath24 and h@xmath0 integration regions were 43204450 and 48005000 for ngc 4593 , and 44204560 and 49205120 for mrk 1511 ( in the observed frame ) .
the resulting light curves include the constant contributions of narrow h@xmath0 and h@xmath24 as well as [ ] @xmath28 blended with h@xmath24 .
light curves for @xmath27 were measured by summation of the best - fitting broad and narrow model components .
this produced a less noisy light curve than the alternate approach of calculating a residual spectrum for each night by subtracting all of the other model components .
however , for ngc 4593 the emission was too weak to produce a useful light curve .
the light curves were computed by summation of the best - fitting model over 44004900 , but any wavelength range would yield the same overall light - curve shape since the model fit assumes a uniform flux scaling factor for the template .
the fc light curves were measured by integration over the same wavelength range used for .
as described above , the residual scatter measured from the [ ] light curves represents a major contribution to the error budget in the light - curve measurements .
therefore , we combine the residual scatter ( 1.2% for ngc 4593 and 2.0% for mrk 1511 ) in quadrature with the measurement uncertainties in the individual light - curve points to obtain a more realistic estimate of the true overall uncertainties .
the emission - line light curves displayed in figure [ figlightcurves ] are shown with these expanded error bars .
addition of the residual flux - scaling scatter to the error budget widens the measurement uncertainties in the cross - correlation lags described below , but only slightly . in figure
[ figlightcurves ] we show only the spectroscopic light curves measured using the v04 template fits because the light curves measured from the bg92 fits are extremely similar in overall appearance .
observations of other agns have shown that the amplitude of variability is typically somewhat smaller than that of h@xmath0 .
for example , in ngc 5548 @xcite found that the variation amplitude was @xmath39%75% of that of h@xmath0 .
to characterize the variability amplitude in ngc 4593 and mrk 1511 , we use @xmath17 as defined in equation 1 , which gives a measure of the normalized rms variability , corrected for measurement uncertainties .
based on the v04 fits , the results for ngc 4593 are @xmath17(h@xmath0)@xmath40 and @xmath17()@xmath41 , while for mrk 1511 we measure @xmath17(h@xmath0)@xmath42 and @xmath17()@xmath43 . with the bg92 template
fits , the @xmath17 values for h@xmath0 are essentially unchanged , while @xmath17 ( ) increases by @xmath44 relative to the values determined from the v04 fits .
thus , both objects show a somewhat higher amplitude of variability in h@xmath0 than , similar to the behavior seen in ngc 5548 @xcite . for agns having a very low starlight fraction ,
it is often possible to measure the agn continuum flux from the spectra directly , without carrying out a decomposition to separate the agn flux from the host galaxy contribution ( e.g. , * ? ? ?
. however , this approach is much less successful for agns having a substantial host galaxy component in the spectra , since the starlight dilutes the variability of the agn continuum and introduces additional random noise due to nightly variations in seeing and target centering within the slit ( e.g. , * ? ? ?
given the substantial contributions of both stellar continuum and in these two objects , the spectral decomposition method is necessary in order to measure the agn continuum flux without contamination or dilution by these other components .
in order to measure the cross - correlation function ( ccf ) for unevenly sampled time series , we employ the interpolation cross - correlation function ( iccf ) methodology and monte carlo error - analysis techniques described by @xcite , @xcite , and @xcite .
emission - line lags for h@xmath0 , h@xmath24 , , and were measured relative to both the @xmath1-band and fc light curves .
ccfs were computed over a temporal range of @xmath45 to @xmath46 days in increments of 0.25 days . for each ccf
we compute two measures of the lag time : @xmath47 , which is the lag at the peak of the ccf , and @xmath2 , the centroid of the ccf for all points above 80% of the peak value @xcite .
table [ xcortable ] lists the lag values .
the quantity @xmath48 listed in table [ xcortable ] gives the peak amplitude of the ccf and is a measure of the significance of the correlation between the two light curves .
we obtain higher @xmath48 values for ngc 4593 than for mrk 1511 , primarily owing to the higher s / n of the ngc 4593 light curves .
lag values are given in the observed frame and can be converted to the agn rest frame by dividing by @xmath49 , where the redshifts are @xmath50 for ngc 4593 and @xmath51 for mrk 1511 .
there are advantages and drawbacks to using either the @xmath1-band data or spectroscopic fc light curves in the cross - correlation analysis .
the @xmath1-band light curves have more frequent sampling than the spectroscopic data , and better temporal sampling will always improve the determination of the ccf .
additionally , the spectroscopic data suffer from systematic effects including variable slit losses due to miscentering and differential atmospheric refraction , while the photometric data are not susceptible to these problems .
however , the @xmath1-band data include contributions of emission - line flux in addition to the agn continuum , while the spectroscopic decompositions can be used to produce a `` pure '' fc light curve without contamination by strong emission lines .
a disadvantage of using the spectroscopic data in cross - correlation measurements is the effect of correlated flux - calibration errors between the emission - line and continuum light curves , which can introduce a spurious signal at zero lag .
we find that using the @xmath1-band light curves produces consistently higher - quality results , but the reverberation lags are consistent when measured against either the @xmath1 band or fc light curves . for completeness , table [ xcortable ]
presents all of the lag measurements based on the spectral decompositions using both the v04 and bg92 templates , and for cross - correlations computed with both the @xmath1-band and fc light curves .
we tested the effect of detrending the data by subtracting a linear fit to the light curves before computing the ccfs .
as described by @xcite , removing long - term secular variations from light curves can often improve the accuracy and significance of cross - correlation results . for mrk 1511 ,
detrending had almost no effect on the ccfs , because the light curves are not characterized by any long - term brightening or fading .
ngc 4593 does show a net increase in luminosity over the duration of our monitoring program , but ccfs computed using detrended data were actually of poorer quality than the original ccfs , having values of @xmath48 that were lower by @xmath52 .
therefore , we chose not to use detrended light curves for our final measurements .
we first describe the results based on light curves measured from the v04 template fits .
comparison with bg92 template fits is presented in section [ sec : templates ] .
figure [ figccf ] illustrates the ccfs measured relative to the @xmath1-band data for emission - line light curves measured from the v04 template fits , as well as the autocorrelation function ( acf ) of the @xmath1-band light curve . for both agns
the ccf has a shape roughly similar to that of the h@xmath0 ccf , but systematically shifted toward longer lags .
lccc|ccc[t ! ] ngc 4593 : & & & & + h@xmath24 vs. @xmath1 & 0.95 & @xmath53 & @xmath54 & 0.94 & @xmath55 & @xmath54 + h@xmath24 vs. fc & 0.94 & @xmath56 & @xmath57 & 0.80 & @xmath58 & @xmath59 + h@xmath0 vs. @xmath1 & 0.92 & @xmath60 & @xmath61 & 0.94 & @xmath62 & @xmath63 + h@xmath0 vs. fc & 0.92 & @xmath64 & @xmath65 & 0.82 & @xmath66 & @xmath67 + vs. @xmath1 & 0.88 & @xmath68 & @xmath69 & 0.85 & @xmath70 & @xmath71 + vs. fc & 0.85 & @xmath72 & @xmath73 & 0.82 & @xmath74 & @xmath75 + mrk 1511 : & & & & + vs. @xmath1 & 0.84 & @xmath76 & @xmath77 & 0.83 & @xmath78 & @xmath79 + vs. fc & 0.81 & @xmath80 & @xmath81 & 0.79 & @xmath82 & @xmath83 + h@xmath24 vs. @xmath1 & 0.91 & @xmath84 & @xmath85 & 0.91 & @xmath86 & @xmath87 + h@xmath24 vs. fc & 0.83 & @xmath88 & @xmath89 & 0.80 & @xmath90 & @xmath91 + h@xmath0 vs. @xmath1 & 0.89 & @xmath92 & @xmath93 & 0.88 & @xmath94 & @xmath95 + h@xmath0 vs. fc & 0.82 & @xmath96 & @xmath97 & 0.81 & @xmath98 & @xmath99 + vs. @xmath1 & 0.80 & @xmath100 & @xmath101 & 0.81 & @xmath102 & @xmath103 + vs. fc & 0.75 & @xmath104 & @xmath105 & 0.73 & @xmath106 & @xmath107 + [ xcortable ] normally , the iccf method computes two versions of the ccf : first by interpolating the continuum light curve ( shifted by each trial value of the lag ) to the temporal steps of the emission - line light curve , and then by interpolating the emission - line light curve to the time sampling of the continuum light curve . with high - quality data ,
the two resulting ccfs are normally very nearly identical , and the final ccf is taken to be the mean of the two @xcite . however , when the two light curves have significantly different temporal sampling frequency , or when the light curves have poor s / n , the two interpolated ccfs can sometimes differ significantly in shape , and occasionally one of the two ccfs may contain features that are clearly spurious while the other has a more regular appearance .
in such a situation it may be preferable to use just one of the two interpolated ccfs rather than taking their mean . for ngc 4593 ,
the two interpolated ccfs are nearly identical , and we base our measurements on the mean ccfs as illustrated in figure [ figccf ] . however , for mrk 1511 , we found that the second version of the ccf ( i.e. , from interpolating the light curve to the time steps of the @xmath1 band ) had a large and unphysical `` notch '' appearing just at the ccf peak .
the other interpolation produced a smooth and regular ccf .
overall the two interpolated ccfs had nearly identical shapes and widths , and they show the same overall lag response aside from the notch appearing at the peak of the first ccf .
the presence of this notch in the mean ccf rendered the measurement of @xmath47 and @xmath2 suspect . since the first interpolated ccf results in a smoothly shaped peak , for mrk 1511 we use only the first interpolated version of the ccf to calculate @xmath47 and @xmath2 , rather than taking the mean of the two interpolations .
this ccf is shown in figure [ figccf ] . days .
this is consistent with the value listed in table [ xcortable ] but with a larger uncertainty due to the peculiar shape of the peak region of the mean ccf .
the value of @xmath2(h@xmath0 ) is nearly unchanged if we use the mean ccf instead of the first interpolated version . ]
the h@xmath0 lags measured relative to the @xmath1-band light curves are in the range of @xmath1084 days for ngc 4593 and @xmath1097 days for mrk 1511 . due to asymmetry in the ccfs , the values of @xmath2 and @xmath47 differ slightly , but not by a significant amount considering the uncertainties in the lag measurements .
the observed trend of h@xmath0 having a longer lag than h@xmath24 is expected based on previous measurements ( e.g. , * ? ? ?
similarly , the fact that the lag is unresolved in mrk 1511 is consistent with expectations based on previous measurements . in agns having h@xmath0 lags of @xmath110 days ,
the lag time is usually undetectable in datasets with nightly sampling ( e.g. , * ? ? ?
* ; * ? ? ?
the emission has a longer lag than h@xmath0 for both agns . in both cases
the cross - correlation between the and @xmath1-band light curves produces a significant ccf peak , with @xmath111 for ngc 4593 and 0.80 for mrk 1511 .
we also measured the emission - line lags relative to the agn fc component light curves .
the resulting @xmath2 and @xmath47 values are generally consistent , within the uncertainties , with the lag values measured against the @xmath1-band data .
however , the ccfs measured using the fc light curves have lower peak amplitude @xmath48 than the corresponding ccfs measured against the @xmath1 band , which we attribute to the poorer temporal sampling and greater calibration uncertainties of the spectroscopic data .
thus , we consider measurements relative to the @xmath1 band to be the best determinations of the emission - line lags , while the lags measured relative to the fc serve as a useful consistency check . as another check
, we measured the fc light curves over a different wavelength range ( 50505150 rest wavelength ) and measured the cross - correlations of the h@xmath0 light curves relative to this revised fc light curve .
the measured lags only changed by a negligible amount ( much smaller than the 1@xmath112 uncertainties ) in comparison with the lags measured to the fc light curves as listed in table [ xcortable ] .
figure [ figccf - fc ] shows the vs. fc ccfs for both sets of template fits . for mrk 1511
, the choice of template has very little impact on the ccf shape or the lag in this measurement .
however , the difference between the two templates is significant for ngc 4593 , where the light curves measured from the bg92 template produce a very jagged and irregular ccf and yield very unreliable lag measurements with large uncertainties on @xmath2 and @xmath47 .
the v04 template , on the other hand , gives a better - behaved result and smaller lag uncertainties .
the cause of this difference is not obvious , but it does appear that the v04 template gives a clearly superior result at least for ngc 4593 .
an additional possible concern is degeneracy between the stellar and agn continuum components in the fit . for both objects
, the light curves of the starlight component do not show any significant time - dependent features other than random noise .
we measured cross - correlations between the stellar continuum component and the @xmath1-band light curves in order to test whether the stellar continuum light curves might contain some residual agn flux .
the resulting ccfs had very low peak amplitudes of @xmath113 and 0.43 for ngc 4593 and mrk 1511 , respectively , and had such irregular structure that no clear lag signal could be measured in either case .
this provides further evidence that the spectral decompositions are properly separating the contributions of the continuum components .
a small bias in reverberation lag measurements can occur when broad - band photometry is used for the continuum flux measurement , due to the contribution of emission - line flux to the photometric light curves .
the passband of the @xmath1 filter includes flux from both h@xmath0 and , both of which lag the continuum variations .
when cross - correlating spectroscopic emission - line light curves against the @xmath1-band light curves , this emission - line contamination in the @xmath1-band data will tend to bias the cross - correlation lag toward values that are lower than the true lag @xcite . to determine the level of this contamination for mrk 1511 as an example , we extrapolated our mean - spectrum fit over a broader wavelength range ( up to 6500 ) and carried out synthetic @xmath1-band photometry on the individual model components . in the mean spectrum ,
the ratios of emission - line to agn featureless continuum ( fc ) flux as measured through a @xmath1-band filter are @xmath114(h@xmath0)/@xmath114(fc ) @xmath115 , and @xmath114()/@xmath114(fc ) @xmath116 .
we simulated the effect of the emission - line flux in the @xmath1-band light curves by creating mock reverberation datasets having properties similar to those of our observed data . following the method of @xcite
, we began by generating simulated agn continuum light curves on a finely sampled temporal grid with bin size equal to 0.01 day , based on a power - density spectrum of the form @xmath117 ( similar to slopes measured from _
agn light curves by * ? ? ?
each light curve was generated over a total duration of 1000 days .
a shorter segment of length 100 days was then selected randomly and normalized to have an rms variability equal to 15% of the median flux .
the h@xmath0 and lines were assumed to respond linearly to continuum variations , with a delta function as the transfer function for simplicity , and with lag times of 6 days for h@xmath0 and 9 days for .
variability amplitudes for the emission lines were normalized so that h@xmath0 had the same 15% rms variability as the continuum , while was set to have a lower rms variability amplitude of 12% .
then , a `` contaminated '' @xmath1-band light curve was created by adding scaled versions of the h@xmath0 and light curves to the fc light curve , in the relative proportions listed above for mrk 1511 .
all of the light curves ( fc , @xmath1 band , h@xmath0 , and ) were then degraded to approximate the sampling and s / n of real data .
from the finely sampled light curves , one flux point was chosen per night , randomly sampled from a window of @xmath118 hours relative to a uniform 24-hour cadence , and over a total spectroscopic monitoring duration of 80 days .
weather losses were approximated by randomly removing 10% of the points from the fc and @xmath1-band light curves and 40% of the points from the emission - line light curves .
random gaussian noise was added to attain s / n = 100 for the fc and @xmath1-band data , and s /
n = 50 for the emission - line light curves .
the light curve was cross - correlated against both the pure fc light curve and the contaminated @xmath1-band light curve , to determine the typical level of the bias , and this procedure was repeated @xmath119 times to build up a distribution of lag measurements for different initial realizations of the light - curve shape .
we compiled the values of @xmath2 for the subset of simulations that produced `` successful '' measurements of lag , meaning that the ccf had @xmath120 and yielded a nonnegative value of @xmath2 ; by these criteria 83% of the simulations were successful .
for this subset , the median lag of relative to the pure agn continuum is @xmath121 days , which is closely consistent with the input lag of 9 days .
( the uncertainty of @xmath122 days represents the 68% confidence interval on @xmath2 for the set of successful simulations .
) when the simulated light curves are cross - correlated against the simulated @xmath1-band data including emission - line contributions , we find a median lag of @xmath123 days .
as expected , this is shorter than the lag measured with respect to the pure agn continuum , but the difference of 0.7 days in the median measurements is smaller than the uncertainty in the measured lag for either of our targets .
thus , the impact of this bias on the emission - line lag measurements is fairly small , most likely below the level of the @xmath124 uncertainties in the lags
. the effect of emission - line contamination in the @xmath1-band light curves would be similar for all of the cross - correlation measurements for a given agn , so the inferred relative sizes of the and h@xmath0 emission regions would be only modestly affected . among the reverberation measurements listed in table [ xcortable ] , the strongest differences between results based on the v04 and bg92 template fits are in the lag of relative to the @xmath1 band .
the bg92 fits give @xmath2 values that are 23 days lower than the values measured from the v04 fits .
this difference appears to stem from the different proportions of fc and fluxes for the two sets of spectral fits .
as can be seen in figures [ figmean - veron ] and [ figmean - boroson ] , the broadened v04 template approaches zero flux density for the emission in the blue wing of h@xmath0 , as well as at the blue end of our fitting region . the broadened bg92 template ,
on the other hand , does not reach zero flux density anywhere in our fitting region , producing a higher `` pseudo - continuum '' level from blended lines and forcing the fc component to a lower overall flux than in the v04 fits .
we suspect that the bg92 template fits are including a small amount of agn continuum flux into the fit component .
this small contribution of agn continuum flux in the light curves would add a spurious signal at zero lag to the cross - correlations between and the @xmath1-band data , slightly biasing the lags toward low values .
these differences stem from the methods used to construct the templates .
the bg92 template is based on an observed spectrum of i zw 1 .
emission lines from several transitions other than were removed , and the continuum was removed by fitting a polynomial to regions between strong emission lines , leaving an approximate emission spectrum .
the v04 template is also based on an observed spectrum of i zw 1 , but was constructed by removing a much larger list of non- emission features , and also removing narrow permitted and forbidden lines which were found to be much stronger in i zw 1 than in most typical seyfert 1 spectra .
each line in the remaining spectrum was then modeled with a lorentzian profile of fwhm@xmath125 km s@xmath12 , producing a noise - free template .
the more rigorous removal of non- emission features and narrow lines by v04 is a major reason to prefer the v04 template over the bg92 template for these fits .
furthermore , the empirical and somewhat subjective procedure used to remove the agn continuum by bg92 may be responsible for the fitting degeneracy between the and fc components described above .
if the agn continuum were undersubtracted when constructing the bg92 template , there would be almost no noticeable impact on the quality of individual spectral fits that used the template , but the mixing of fc and emission would likely bias the reverberation lags measured for the component . for ngc 4593 , the ccf measured between the and fc light curves is certainly better when the v04 template is used , compared with the bg92 template .
the reason for this difference is not clear , and it only appears to occur for ngc 4593 and not mrk 1511 , but it does add one additional reason to prefer the v04 template fits . for all of these reasons , we consider the measurements done using the v04 template fits to be our best - quality results .
most importantly , our main results are independent of the template choice : in either case we find that emission does show a strong reverberation response , with a lag time longer than that of h@xmath0 . as described in the appendix , we find similar results when using the newly released multi - component templates of @xcite . with these new templates ,
we obtain slightly different values of @xmath2 for the emission lines , but the lags are still consistently longer than those of h@xmath0 , with @xmath2()/@xmath2(h@xmath0)@xmath126 .
this further confirms that the larger size of the -emitting region ( compared with h@xmath0 ) is genuine and not an artifact of template choice .
it is also worth noting that the different template fits yield differing amounts of and emission underlying the broad h@xmath0 profile .
the uncertain amount of blending of h@xmath0 with other emission features ( each having a different lag relative to the continuum ) could represent a limiting factor for the accuracy of high - fidelity reverberation measurements , particularly for velocity - resolved lag measurements which attempt to determine the distribution of lag across the velocity width of the line . any ambiguity in decomposing the h@xmath0 spectral region into different emission - line components
should be considered as a source of systematic uncertainty in determining the h@xmath0 lag distribution across the line profile .
this is the first time that such clear reverberation signals have been seen for the optical blends in seyfert galaxies .
our high - cadence monitoring data reveal that the emission in these galaxies does reverberate on short timescales in response to continuum variations , with a well - defined cross - correlation lag time .
this gives direct evidence for an origin of the emission in photoionized gas in the blr .
the ambiguity in previous reverberation results for ngc 5548 @xcite and ark 120 @xcite might be attributable to the lower cadence of the monitoring data used in these studies , although in both cases the monitoring duration spanned several years .
it is also possible that the variability behavior in our two objects is not representative of the entire population of seyferts .
measurement of variability over a broad range of agn properties should be a high priority for future high - cadence reverberation - mapping programs . by exploring a broad range of luminosities
, it might be possible to test whether the optical emission follows a radius - luminosity relationship similar to that of the h@xmath0 line , with a luminosity dependence of approximately @xmath127 @xcite
. it would be particularly interesting to test whether the ratio of to h@xmath0 radii varies systematically along the eigenvector 1 sequence as a function of @xmath8 .
if the -emitting gas is infalling toward the central engine , as proposed by @xcite , then there might also be observable correlations between the reverberation lag and the redshift of the lines , although evidence for radial inflow remains controversial @xcite .
the measured values of @xmath2 give an approximate mean radius for the zone from which each line is emitted within the blr .
comparing the lags of and h@xmath0 for the preferred v04 template fits , we find @xmath3()@xmath128(h@xmath0 ) @xmath129 and @xmath5 in ngc 4593 and mrk 1511 , respectively .
this gives a direct indication that the emission arises predominantly in the outer portion of the blr on larger scales than the h@xmath0 emission region .
this conclusion is consistent with several other recent lines of evidence for an outer blr location for the emission @xcite .
we use the term `` outer blr '' to denote emission from a region having a larger mean size than the h@xmath0 emission zone . aside from ,
the only broad emission lines seen to have lags longer than that of h@xmath0 are ] @xmath130 @xcite and h@xmath6 ( e.g. , * ? ? ?
* ; * ? ? ?
thus , the emission regions for these lines correspond to the outermost observable portion of the blr .
the bg92 template fits give different ratios of to h@xmath0 lag , but we find that @xmath2 ( ) @xmath131 @xmath2(h@xmath0 ) for either template .
the longer lag of relative to h@xmath0 does not mean that the h@xmath0 and emission regions are physically distinct ; in fact , there must be a very substantial radial overlap between them .
recent progress in transfer - function modeling for reverberation - mapping data has provided direct illustrations of the radial extent of the blr as seen in balmer lines . for arp 151 ,
the transfer functions show that the h@xmath0 emission extends over a broad radial zone with an outer extent that is more than twice as large as @xmath132 , and the h@xmath6-emitting region is several times larger in extent than the mean h@xmath0 radius @xcite .
if the ratios of to h@xmath0 size for ngc 4593 and mrk 1511 are typical , then the -emitting zone would encompass the outer portion of the h@xmath0-emitting zone of the blr and beyond , probably corresponding roughly to the region emitting h@xmath6 .
this echoes the recent finding by @xcite that quasar h@xmath0 profiles generically contain an intermediate - width component which may be emitted from the same region that produces , while the very broad component of h@xmath0 would be emitted from smaller radii .
further support for a link between the region and the outer portion of the balmer - line emitting zone of the blr comes from a new principal - component analysis of quasar spectra by @xcite .
they demonstrate that one of the primary eigenspectra of their quasar sample is essentially the sum of two components : the emission spectrum and an intermediate - width core component of the balmer lines , which are readily interpreted as arising from a similar spatial region .
@xcite found that line widths in sdss quasars are typically about 3/4 of the h@xmath0 widths . in order to test whether ngc 4593 and mrk
1511 are consistent with this trend , we examine the results of the spectral fits done using the bg92 template , since this closely follows the fitting method used by @xcite .
we assume that the total velocity width of fwhm is given by the quadrature sum of the fwhm of the broadening kernel from the fit and the fwhm of the line profiles in the template itself which we take to be 900 km s@xmath12 for consistency with @xcite .
we also correct the observed widths for an instrumental broadening of fwhm@xmath133 km s@xmath12 @xcite .
then , we obtain widths of fwhm@xmath134 km s@xmath12 for ngc 4593 and @xmath135 km s@xmath12 for mrk 1511 , where the uncertainties are based on the night - to - night scatter in the width of the gaussian broadening kernel determined by the fitting procedure
. the broad - h@xmath0 component models for the two agns have fwhm = @xmath136 km s@xmath12 for ngc 4593 and @xmath137 km s@xmath12 for mrk 1511 . in both cases , then , fwhm()/fwhm(h@xmath0 ) is almost precisely 0.75 , closely consistent with the average result for the @xcite sample .
however , we also find that the inferred widths seem to be substantially dependent on choice of template .
using our results from fitting with the v04 template , we obtain broader widths of fwhm = @xmath138 km s@xmath12 and @xmath139 km s@xmath12 for ngc 4593 and mrk 1511 , in both cases broader than the h@xmath0 fwhm values .
this appears to be the result of a combination of factors including different relative strengths for individual lines between the two templates , and the fact that the v04 template is constructed using lorentzian models fitted to each line .
one consequence of these lorentzian profiles is that the fwhm values of the template itself and of the gaussian broadening kernel do not simply add in quadrature to give the fwhm of the total line profile ; the fwhm of the gaussian - broadened template profile is broader than the quadrature sum of the template fwhm and broadening kernel fwhm .
based on the fits done using the bg92 template , we can conclude that these two agns have fwhm()/fwhm(h@xmath0 ) ratios consistent with the mean value found by @xcite for a large sdss sample , so there is no evidence that they are outliers from the normal agn population .
however , the absolute determination of width seems to be subject to substantial systematic uncertainty due to the details of how the fit is performed and choice of template .
further investigation of the cause of this discrepancy in widths for different templates is beyond the scope of this paper , but this problem is likely to affect all inferences about profile widths measured by template fitting , particularly when the individual features are completely blended into a pseudocontinuum as is the case in these two objects .
in addition to the systematic issues related to choice of template , there are some additional and important caveats to note about the measured lags .
the ccf methodology only gives a single , simplistic measure of the lag time for an emission line , whereas in reality a given spectral line will be emitted over a broad range of radii .
it might be possible to obtain more detailed information on the radial distribution of emissivity by applying the geometric modeling methods described by @xcite or by applying techniques to extract the shape of the transfer function from the data @xcite .
however , the light curves are relatively noisy and higher - quality data might be required , perhaps over a longer monitoring duration , in order to go beyond the simple cross - correlation determination of the lag presented here . furthermore , our spectral fits assume a uniform flux scaling for the entire template , which results in an average lag measurement for the entire complex of blends , but this method is unable to explore the possibility of different response times for different lines .
recent work has improved on traditional template - fitting methods by allowing for different behavior among different groups of lines @xcite , and @xcite demonstrated that in ark 564 , lines from different multiplets showed differing levels of correlation with continuum variations .
application of such methods to reverberation - mapping data could potentially detect or constrain differences in reverberation timescale for different multiplets as well .
the primary conclusions of this study are that measurement of the reverberation lag of the optical blends is indeed possible in favorable cases , and that in these two agns the emission responds directly to continuum variations with a lag time that corresponds to the outer portion of the blr , somewhat larger than the h@xmath0-emitting radius . with well - sampled data it is possible to detect this reverberation signature with a level of significance comparable to that of high - quality h@xmath0 reverberation measurements .
detection of reverberation does require some special circumstances , in particular a high amplitude of continuum variability , as well as relatively strong emission overall . in all other objects from our 2011 sample ,
either the emission was too weak ( as in mrk 50 ; * ? ? ? * ) , or the overall flux variability was too low for measurements like these to be successful .
finally , these results illustrate the value of carrying out spectral decompositions of reverberation - mapping data , in order to measure accurate light curves for , , and other weak or low - amplitude spectral features .
application of these methods for measurement of emission - line light curves for our entire 2011 sample will be described in future papers in this series .
we are extremely grateful to the lick observatory staff for their outstanding assistance during our 2011 observing run .
the lick agn monitoring project 2011 is supported by nsf grants ast-110812 , 1107865 , 1108665 , and 1108835 .
a.p . acknowledges support from the nsf through the graduate research fellowship program .
a.v.f.s group at uc berkeley received additional funding through nsf grant ast-1211916 , gary & cynthia bengier , the richard & rhoda goldman fund , the tabasgo foundation , and the christopher r. redlich fund . kait and its ongoing operation were made possible by donations from sun microsystems , inc .
, the hewlett - packard company , autoscope corporation , lick observatory , the nsf , the university of california , the sylvia & jim katzman foundation , and the tabasgo foundation . t.t . acknowledges a packard research fellowship . the work of d.s . and r.j.a . was carried out at jet propulsion laboratory , california institute of technology , under a contract with nasa .
research by j.l.w . is supported by nsf grant ast-1102845 .
j.h.w . acknowledges support by the national research foundation of korea ( nrf ) grant funded by the korea government ( mest ) ( no .
2012 - 006087 ) .
the west mountain observatory receives support from nsf grant ast-0618209 .
we thank the anonymous referee for helpful suggestions .
we mourn the tragic passing of our friend and collaborator , weidong li , who devotedly oversaw the nightly operation of kait and taught us much about photometry .
this work is dedicated to the memory of lick observatory staff member greg sulger .
after this paper was submitted , the referee informed us that a public release of the multicomponent templates from @xcite had recently been announced @xcite .
although a full study of variability using these new templates is beyond the scope of this paper , we carried out an initial examination by adapting our fitting method to use these templates .
as described by @xcite and @xcite , the template set includes spectra describing lines from four separate multiplet groups ( denoted as the _ f _ , _ g _ , _ s _ , and _ p _ groups ) , as well as an additional template containing other lines found in the
i zw 1 spectrum .
carrying out fits using this set of five templates gives substantially more freedom to accurately fit observed spectra than the monolithic templates of bg92 or v04 , at the cost of adding four additional free parameters to allow for the individual flux scaling of each component .
figure [ figmean - kov ] illustrates the fits to the mean spectra using these multicomponent templates .
similar to the fits with the v04 and bg92 templates , the model reproduces the overall spectral shape well , but there are significant differences in the fit details compared with the monolithic templates .
using the @xcite templates , the fit assigns relatively more flux to the starlight component and less to the fc . also , these fits force the fluxes of all three lines to zero , and the red shelf of h@xmath0 becomes dominated by , similar to the results from using the bg92 template .
another notable difference is seen at the shortest wavelengths , below about @xmath140 . in this region , the primary contribution is from the _ p _ group , and the fit forces the normalization of this component to zero .
this is most likely a spurious result due to degeneracies in the model fitting process , but it occurs consistently when fitting each individual spectrum for both agns . these multicomponent templates are optimally suited for use with agn - dominated spectra having very strong emission , such as the objects studied by @xcite and @xcite , but the high starlight fraction and relatively weaker emission in our targets presents a more difficult case study for constraining the weights of the five - component model . in order to check for differences with respect to the previous spectral fits , we measured h@xmath0 and light curves based on these decompositions following the same methods described previously , and carried out cross - correlations of these light curves against the @xmath1-band continuum .
the light curve was integrated over the same spectral region used for the monolithic templates , in this case corresponding primarily to flux from the _ f _ group template . for ngc 4593
, we find @xmath2(h@xmath0)@xmath141 days , and @xmath2()@xmath142 days . for mrk 1511 , the results are @xmath2(h@xmath0)@xmath143 days and @xmath2()@xmath144 days .
these results are generally consistent with the measurements done using the bg92 and v04 template fits , with the largest disagreements only being at slightly greater than the @xmath124 level .
the modest disagreements further highlight the fact that the reverberation results are somewhat sensitive to the different structures of the templates . despite these differences ,
our primary result is essentially unchanged : the reverberation lags based on the new template fits are @xmath145 longer than those of h@xmath0 , pointing to an origin for the emission in the outer portion of the blr . in future work , these multicomponent templates may prove to be most advantageous when fitting very high s / n spectra of agns having smaller starlight contributions and stronger lines whose relative amplitudes can be more tightly constrained in the fits . | the prominent broad emission blends in the spectra of active galactic nuclei have been shown to vary in response to continuum variations , but past attempts to measure the reverberation lag time of the optical lines have met with only limited success . here
we report the detection of reverberation in two seyfert 1 galaxies , ngc 4593 and mrk 1511 , based on data from a program carried out at lick observatory in spring 2011 .
light curves for emission lines including h@xmath0 and were measured by applying a fitting routine to decompose the spectra into several continuum and emission - line components , and we use cross - correlation techniques to determine the reverberation lags of the emission lines relative to @xmath1-band light curves . in both cases
the measured lag ( @xmath2 ) of is longer than that of h@xmath0 , although the inferred lags are somewhat sensitive to the choice of template used in the fit . for spectral decompositions done using the template of @xcite , we find @xmath3()/@xmath3(h@xmath0 ) @xmath4 in ngc 4593 and @xmath5 in mrk 1511 .
the detection of highly correlated variations between and continuum emission demonstrates that the emission in these galaxies originates in photoionized gas , located predominantly in the outer portion of the broad - line region . |
the concept of the pseudogap has born from the analysis of hole - doped cuprate superconductor low - energy excitations .
an extensive literature exists concerning this feature and its interpretation [ 1 - 9 ] .
the pg has been identified by various experimental methods and seems to be a characteristic phenomenon of doped cuprate superconductors .
it is nearly material - independent [ 10 ] .
the pg is exposed also in the normal state material at t@xmath1t@xmath0 .
this is in essential contrast with the true strongly material dependent superconducting gap ( scg ) observed at t@xmath2t@xmath0 .
there have been misunderstandings in literature in discussing various gaps in comparable conditions .
the pg can enter the superconductivity dome and be observable in this region as a normal state gap ( nsg ) [ 11 ] .
below t@xmath0 the pg can remain masked by the excitations of superconducting states . at a fixed doping
there is no transformation of the pg into a scg .
the magnitude of the pg is usually characterized by a temperature t@xmath3 which is strongly doping - dependent .
this t@xmath3 line does not represent a real phase transition [ 12 , 13 ] .
the pg can not be directly jointed to the superconducting order through its fluctuations in the normal state .
one has shown that these fluctuations are quenched far below t@xmath3 [ 14 - 16 ] . at present the origin of the pg
is attributed to extrinsic genesis in respect of superconductivity [ 2 , 6 , 14 , 16 - 19 ] .
the formation of the pg can be associated with various types of the orderings of lattice and electronic nature [ 20 - 25 ] , accompanied also with time reversal , spatial rotations and translation symmetry breaking [ 26 , 27 ] .
these orderings can be competitive with superconductivity in distinct momentum regions which otherwise participate in pairing .
then a depletion of excitation density around the fermi energy appears .
the presence of a magnetic counterpart of the charge - channel pg has also been reported [ 1 , 16 , 28 ] .
in general , the role of the spin - subsystem remains in the pg aspect not well understood .
both the pg and the scg appear as quasiparticle - type excitations .
however , the scg is of coherent- and the pg of noncoherent nature excitation [ 26 , 29 , 30 ] .
these gaps can represent the actions of different momentum space regions at the same dopings .
the cuprate superconductivity is initiated by a necessary doping process .
the first major result of this procedure is the metallization of the sample .
it means that the chemical potential ( @xmath4 ) must lie in a partially filled band of the doping influenced spectrum .
the basic ( af ) order of cuprate becomes destroyed .
the distortions of the crystal interior , which create the pg in the spectrum of the doped material , appear .
this process at very low dopings is better known for hole - doped samples .
it starts by forming hole - centered ferrons [ 31 ] on af networks [ 32 , 33 ] .
an extended doping stimulates the percolation of such mesoscopic dopants - associated cuo@xmath5 plane plaquettes [ 32 ] .
a recent paper [ 34 ] has followed on the atomic scale the evolution of the electron structure from an insulator to superconductor , passing some kind of weak insulator state .
the pg becomes built up on the background of the weakening of af correlations with an extended doping . for electron - doped systems ,
where the af and superconducting domains overlap markedly , the problem needs further investigations .
the pg and the scg can be observed in the same underdoped region .
the energy scales for these two types of gaps are here rather different , as also their development with doping are opposite .
the pg diminishes , the scg grows with doping .
this is the simplest indication to differentiate them .
coming nearer to the critical point ( cp ) on the phase diagram , the mentioned mismatch in energy is vashed out .
the phase diagram cp is defined as the doping concentration , where the nsg becomes closed [ 28 , 35 - 37 ] .
fermi surface reconstruction appears here . behind the cp the mixed carrier system behaves more as a fermi liquid .
the pg appears as being selective in respect of momentum space regions , where it is created and observed in the corresponding spectral windows . in hole - doped superconductors
the pg is merely an
antinodal ( ( @xmath6)-type brillouin zone centered regions )
the scg at underdoping is of ( @xmath7 ) type nodal phenomenon . on the route between these points
the pg is depressed .
the carriers in the regions where the pg is formed must behave as insulating below t@xmath3 . when passing this curve insulator to metal transitions in the momentum - restricted region are expected and observed [ 38 - 40 ] .
cuprates are electronically highly correlated materials .
the reconstruction of the crystal interior under doping must be taken into account .
the mott - hubbard type af spectrum of a pure sample will be reorganized to form the superconducting playground .
this is the second essential result due to doping .
the perturbed spectrum includes a qualitatively essential effect .
new electron states ( bands ) inside the charge - transfer gap are created [ 26,41 - 46 ] .
one speaks about the creation of midgap- , polaronic- , or defect bands .
doping can regulate the extension of such states and bring them into overlap with the itinerant bands .
the superconductivity acquires the multiband nature .
the electron spectrum near the fermi surface , being actual for the superconductivity , is different for the hole and electron dopings . in the hole case the region near the top of the ( mainly ) oxygen band between the upper ( uhb ) and lower hubbard ( lhb )
is involved . for
electron doping it is the region near the bottom of the uhb . for superconductivity the part of the elecron spectrum near the fermi surface is acual .
the modifications due to doping with the creation of the new ( defect ) bands appear just here . for hole and electron dopings
these spectral regions are different [ 41 , 42 , 72 ] . as largely known , in the hole case it is the area near the top of the ( mainly ) oxygen band .
this band is positioned between the upper ( uhb ) and lower hubbard ( lhb ) bands of the pure af sample .
for the electron doping the region near the bottom of the uhb has been established to be active . the cuprate superconductivity mechanism remains not well understood at present , c.f .
e.g. [ 47 - 51 , 56 ] . the approach of our group has been based on a multiband nature of the doped cuprate , with the original idea to use the interband pairing interaction between the itinerant and doping - induced bands .
model realizations of such type approaches for hole - doped cuprates have given general conclusions and results in a qualitative agreement with experimental data [ 56,57 ] .
the pg on the phase diagram appears here naturally [ 56 - 62 ] .
the later consideration of electron - doped systems by analogous interband pairing scheme can be followed in [ 56,62 ] .
the global similarity of the phase diagrams for both types of doping arise from the topologically comparable physical approach .
the essential differences between them stem from the perturbation of different atomic sublattices of the material .
recently the pairing glue in the cuprates has been identified as being of electronic origin [ 63 ] .
this justifies the use of interband pairing interaction in our approach ( exchange type pair transfer ) with its outstadning advantages [ 64 ] . in this paper
we analyse the pg - s in electron doped cuprates .
this problem is not very clear - the recent review paper [ 65 ] summarizes that there are substantial signatures of pg effects in the electron - doped compounds . in this aspect
also papers [ 3 , 15 , 66 - 68 ] can be consulted and theoretical approaches [ 69 - 73 ] must be mentioned . our original definition [ 55 , 58 , 59 ] of pg excitations exposes the pg as a natural phenomenon on the multiband background [ 74 ] .
it is based on the possibility of the appearance of bands intersected or not intersected by the chemical potential ( @xmath4 ) .
namely , the pg is considered as the minimal quasiparticle excitation energy of a band ( @xmath8 ) not containing @xmath4 @xmath9 it is supposed that @xmath10 . in ( 1 ) @xmath11 stands for the band energy and @xmath12 for the scg gap of the same band . in the case of a nonrigid spectrum @xmath4 can be shifted into @xmath13 , so that @xmath14
. then the normal state contribution into @xmath15 vanishes .
one reaches a phase diagram cp . behind a cp
the @xmath16 remains represented by the sole scg @xmath17 .
the reorganizations in the multiband spectrum by doping , or , say , by pressure can influence the pg until its disappearance .
the changes in band overlaps become of first significance in the problem of the pg .
when the normal state contribution prevails , the pg will be large and repeats practically the behaviour of the nsg . in the neighbourhood of
the cp @xmath4 enters the given band ( or the band reaches @xmath4 ) .
the normal state contribution into the pg will be small and both components of @xmath18 can show a comparable magnitude .
the quenching of superconductivity ( @xmath19 ) can then extract the nsg in this spectral region .
basing on versatile knowledge , a 4-band model of the active part of electron - doped cuprate spectrum ( fig . 1 ) has been proposed [ 62 ] .
there is the uhb (@xmath20 which extends up from its antinodal ( an ) ( @xmath21-type bottom at @xmath22 .
the lhb-(@xmath23 is situated deep and has a nodal ( n ) @xmath24 -type top .
both these bands are of width @xmath25 .
midgap bands appear under the @xmath26-band bottom on the account of both hubbard bands . to differentiate the functionality of the defect states in the momentum space
we introduce two subbands . the @xmath27 one is an centered with the bottom lying at d. the @xmath28 one is n - centered with the bottom taken as energy zero .
these bands arise from the perturbed copper sublattice and develop with the destruction of the af order , cf . also the recent calculation [ 25 ] .
the total number of states ( 4 bands ) is normalized to 1 .
the concentration of doped electrons for one cu site per one spin is c. the undoped material ( c=0 ) has @xmath29-weights of states @xmath30 .
electron - doping exhausts them to ( 1/2-c ) .
the @xmath8-bands are both of weight c. there is a special spectral weight transfer from the hubbard components to doping - induced midgap states
. extended doping shifts the midgap bands towards the @xmath26-band bottom and the bare gaps between them become removed . this well - known trend can be taken into account by writing for the top energies of the @xmath27- and @xmath28-bands @xmath31 and @xmath32 correspondingly .
this spectral behaviour is decisive for the behaviour of the pg - s on the phase diagram . as a result ,
the progressive doping leads to the formation of a common carrier family .
the constant densities of states read @xmath33 ; @xmath34 , i.e. the midgap bands grow while the hubbard spectrum tends to total ruination at @xmath35 .
the chemical potential is shifted upward with electron doping starting from a position inside the midgap states as expected , e.g. [ 65 ] .
one finds the first cp on this route to be reached at @xmath36 . here
the first defect - uhb ( @xmath37 ) gap becomes closed .
one finds for @xmath38 @xmath39 which lies on the higher midgap subband top ( @xmath28 remains filled ) .
it is known [ 15 ] that at low dopings the electrons occupy ( @xmath40-neighbourhood and extending fermi surface pockets are formed . for @xmath41 the upper midgap band and the uhb overlap and
define a common ( @xmath42 @xmath43 the second cp ( c@xmath44 appears when the nodal midgap states reach the @xmath45 mixture .
then the disposition of bands and @xmath46 tend to maximize t@xmath47 .
so , at @xmath48 where @xmath49 , @xmath50 one really observes that now new spectral intensity around @xmath51 emerges to the game [ 75,78 ] .
this can be attributed to hole - like fermi surface pockets formation [ 78 ] .
the joint smooth rising up @xmath46-curve can be found in [ 62 ] .
note that the positions of the phase diagram cp - s are essentially determined by the changes in the bands overlap .
the same concerns the natural reconstruction of the fermi surface with doping .
there are two pg - s in our model connected to the bands @xmath26 ( uhb ) and the defect @xmath28 not intersected by @xmath46 at low dopings @xmath52 .
the antinodal uhb pg vanishes as the first cp is reached . at the same time
, the nodal - centered pg has a constant contribution corresponding to the energy difference of both @xmath53-bands .
one has for @xmath52 @xmath54^{1/2},\ ] ] @xmath55^{1/2}.\ ] ] at @xmath56 the energy separation of occupied and empty states is removed .
this points to the vanishing long - range af order .
the calculation of the magnetic susceptibility in a comparable case has shown [ 75 ] that deviations from the half - filling cause rapid reduction of the spin correlation length .
for @xmath57 the chemical potential becomes changed to @xmath58(eq.3 ) as @xmath27- and @xmath26-bands overlap .
the smaller pg exhibits now an exposed dependence on doping @xmath59^{1/2}.\ ] ] this gap closes at @xmath60 in the normal state .
there remain no nsg near the uhb bottom .
one expects that here the tracks of the ( short - range ) af correlations , which survived until @xmath60 will be quenched .
the af order vanishes now at all . as a result ,
two pg - s , having their own relation to the af order , are expected for electron - doped cuprate superconductors .
however , the different assignment of ( @xmath61 ) generic momentum space points must be held in mind .
the presence of two pg - s in r@xmath62ce@xmath63cuo@xmath64 seems to follow from the data of [ 66 ] , see also [ 65 ] .
the appearance of two pg - s in our model is the result of the supposition that the defect ( @xmath65 ) and ( @xmath66 ) centered regions of the momentum space are created with a gap between them . without this assumption
there would be only one scg and one pg present .
the mean field hamiltonian for the proposed 3-band model of the active part includes the uhb and @xmath67 , @xmath68 defect bands .
@xmath69\ ] ] @xmath70\ ; .\ ] ] here the band energies @xmath71 are counted from the chemical potential .
the lhb remains filled and figures only in the determination of @xmath72 .
the constant interaction strength ( @xmath73 ) of the uhb with both defect bands are taken equal ( so @xmath74 ) .
the integration ( @xmath75 ) for defect bands is performed over the correspoding energy intervals with their densities of states .
the hamiltonian ( 1 ) can be simply diagonalized and the results for superconducting gaps with their definition are given by the following expressions ( @xmath76 ) @xmath77 the quasiparticle energies are of an usual form @xmath78 these expressions are used in writing ( 1 ) and ( 9 ) .
gaps ( 9 ) tend to zero at a common @xmath79 .
the calculated normal state pseudogaps and the t@xmath47 domain for a , , representative electron - doped cuprate are shown in fig.2 .
the following input parameters have been used : @xmath80 ; @xmath81 ; @xmath82 and @xmath83(ev ) . the interband coupling constant w=0.17 ev between the midgap bands and the upper hubbard component guarantees then the maximum @xmath84k at @xmath85 .
the critical points of the model lie at @xmath86 and @xmath87 .
reasonable modifications in this parameter set do not lead to essential changes in the general results , cf . [ 62 ] .
the first larger uhb - associated pg @xmath88 falls to zero at @xmath89 .
this point is close to the onset of superconductivity .
the calculated [ 76 ] plane coherence length starts here a rapid drop , see also [ 75 ] .
the behaviour of the spin stiffness can be followed in [ 77 ] .
one expects near @xmath89 the long - range af correlations being removed .
the second cp @xmath90 marks the disappearance of the smaller pg @xmath91 and the vanishing of the survived short - range af correlations as well , i.e. the af correlations altogether .
this conclusion is to some extent supported by the finding that with approaching the superconducting order ( and vanishing pg ) the af correlations are of a short - range and not of the long - range nature [ 77 ] .
note also that according to [ 78 ] the long - range af order is not necessary for the development of the pg .
the second cp appears at a position inside the superconducting dome before the maximum @xmath79 is reached at optimal doping .
this means that the domains , where ( short - range ) af fluctuations are preserved , overlap the superconductivity region markedly .
this circumstance is usually mentioned [ 64 ] , especially in comparison with the hole - doped case . beyond @xmath60
the low - energy excitation spectrum is expected to be presented by the superconducting gaps @xmath92 and @xmath12 .
the large fermi surface composed of all overlapping bands supports the optimization of @xmath93 .
when analysing the spectral manifestations of multiple pseudo- and superconducting gaps , one must be careful .
the smaller gap can remain hidden under the larger one and it can also mask the sharpness of the larger gap
. there can be problems with the separation of the constant @xmath94 on the background of @xmath95 without considering the angular polarization difference .
the superconducting contribution @xmath12 into @xmath96 becomes remarkable with an extending doping towards @xmath60 .
this is illustrated in fig.3 .
the bands overlap dynamis of the present model allows one to explain some new effects .
this can be illustrated by fig.4 , for example , for @xmath28 and @xmath26 bands being shifted in the momentum space .
namely , the presence of hole- and electron - type carriers has been detected in electron - doped superconductors [ 64,76,77 ] . in the case of the @xmath46 position in fig.4 it intersects both the electron- and hole - like pockets .
analogously , the observation of steep and flat dispersions for carriers [ 81 ] can be explained .
also , one can expect the observation of the , , foreign-type carriers as compared with the ones introduced by doping .
however , there is an essential requirement that the corresponding , , active pg has been closed ( upper band in fig.4 ) . in addition , paper [ 82 ] states that although both hole and electron pockets are known in electron - doped cuprate superconductors , there , , is no evidence so far of electron pockets in hole - doped cuprates .
this finding can be explained by using the mentioned relative behaviour of the pg and the , , foreign carriers effect . for electron - doped materials
it is known , as also found in the present work , that the pg ( curve 3 in fig.2 ) touches the @xmath93 domain at a relatively , , early level .
contrary to the hole - doped materials , the end of the pg is found ( also calculated [ 56 ] ) to extend nearly until the overdoped end of the @xmath93 domain .
the doping interval for the effect under consideration will be narrow .
in general , the observation of `` foreign '' carriers as compared to the ones introduced by doping is expected to start with the vanishing of the corresponding pg .
an extended comparison of the present model , formulated for a , , typical electron - doped cuprate superconductor , with the experimental data is a complicated task .
the mainly investigated systems are of the type r@xmath62m@xmath97cuo@xmath64 ( r rare earth ; m ce or th ) .
the data for a distinct material and ( naturally ) between various systems are dispersive .
it seems that the general results obtained by our model mechanism justify its use in various analyses .
especially the rised problem of the existence of two pg - s in electron - doped cuprates is of interest .
the theoretical approach itself can be improved in various aspects , e.g. taking the pairing strengths for the @xmath98 and @xmath99 channels to be different , or precising the parameter set .
in conclusion , the present model approach suggests the appearance of two pseudogaps in electron - doped cuprates .
these gaps are connected with the destruction of long- and short - range antiferromagnetic correlations .
there are two critical points on the doping phase diagram .
the larger one is centered in the antinodal momentum space region at small dopings .
the second one is expected to be mainly a near - nodal event .
its doping dependence is exposed behind the first critical point .
the second critical point lies before the maximal @xmath93 .
antiferromagnetic and superconducting domains overlap markedly . nowadays
multiband pairing schemes are applied to novel , famous or artificial superconducting systems .
the appearance of a pg - type excitation in the spectra of these materials can be considered , as illustrated also in the present work , to be an indication of a multicomponent ( probably three - band [ 61 ] ) nature of the superconducting playground . on the other hand , in the case of a multicomponent fermi surface intersected simultaneously by the chemical potential
, one can expect high transition temperatures by the functioning of interband coupling channels ( missing pg ) .
the behaviour of the pg @xmath101 ( [ eq1 ] ) near the second critical point @xmath102 . behind @xmath60
it continues as @xmath12 .
for @xmath103 the @xmath12 ( 2 ) is hidden in @xmath104 .
( 3)- the normal state contribution to the pg . | the pseudogap ( pg ) excitations in the framework of multiband superconductivity are analysed .
calculations for the doping phase diagram of electron - doped cuprate superconductors have been made .
a nonrigid multiband model has been used .
the interband pairing acts between the upper hubbard band ( uhb ) and midgap states created ( and shifted upwards ) by doping .
the appearance of two pg and two phase diagram critical points ( cp ) is suggested .
the larger antinodal pg reflects the destruction of the long - range antiferromagnetic ( af ) order .
the second , smaller pg is a near - nodal event with exposed doping dependence behind the first cp .
it reaches the t@xmath0 - dome .
model calculations have been made .
the peculiarities of the phase diagram are determined by the doping - stimulated band overlap dynamics .
the observation of , , foreign carriers as compared to the ones introduced by doping is expected to start with the vanishing pg .
pacs : 74.20.-z ; 74.72.-h ; 74.25.dw , electron - doped cuprates ; superconductivity ; pseudogaps |
in this paper we are interested in periodic solutions of the differential equation @xmath0 where @xmath1 , with @xmath2 , and @xmath3 is a continuous function , @xmath4-periodic in the first variable , with @xmath5 a fixed positive constant .
we are going to study the existence of _ bouncing _ periodic solutions .
in particular we are looking for solutions @xmath6 solving when @xmath7 and satisfying a _ perfect bounce condition_. such a condition describes a perfectly elastic bounce on the sphere @xmath8 : the speed has the same value before and after the bounce but the sign of the radial component changes .
the main results can be applied also to a class of systems with a particle hitting a cylinder @xmath9 . in the case of a proper cylinder ( i.e. for @xmath10 and @xmath11 ) , it models , for example , a particle subject to a periodic central electric field and to an elastic force ( see figure [ cylfig ] ) .
similar situations can be viewed in the case of a proper sphere , for @xmath12 .
we will focus our attention , at first , to the case of spheres , postponing the treatment of the case of cylinders to section [ seccil ] . by the radial symmetry of the equation , every solution of is contained in a plane , so we can pass to polar coordinates and consider solutions to the following system @xmath13 \rho^2 \vartheta ' = l \ , , \end{cases}\ ] ] where @xmath14 and @xmath15 is the angular momentum .
the bounce condition could be easily written in the following way @xmath16 we emphasize that the bounce does not affect the behavior of @xmath17 . in this paper
we are going to study the existence of rotating periodic solutions , performing a certain number @xmath18 of revolutions around the sphere in the time @xmath19 and @xmath4-periodic in the @xmath20 variable , i.e. such that @xmath21 the existence of periodic solutions of large period @xmath19 was previously studied by fonda and toader in @xcite in the setting of a keplerian - like system where the planet is viewed as a point ( see also @xcite for other situations ) .
problems modeling the motion of a particle hitting some surfaces have been widely studied in literature in different situations , see e.g. @xcite .
the simpler system with impacts is given by the so - called impact oscillator ( see e.g. @xcite ) where a particle hits a wall attracted towards it by an elastic force . the existence of bouncing periodic solutions of such systems has been discussed e.g. in @xcite . however , to the best of our knowledge , it seems that similar existence results on rotating periodic solutions with impact on spheres ( or cylinders ) of positive radius have not been presented yet .
let us now explain in details what we mean by the term `` bouncing solution '' , borrowing the definition given by bonheure and fabry in @xcite , which we recall for the reader convenience . in it ,
the constant @xmath22 indicates the @xmath23-coordinate of the _ wall _ against which the solution bounces .
[ bsol ] consider a scalar second order differential equation @xmath24 where @xmath25 is a continuous function .
a @xmath22-_bouncing solution _ is a continuous function @xmath26 , defined on some interval @xmath27 , such that @xmath28 for every @xmath29 , satisfying the following properties : * if @xmath30 is such that @xmath31 , then @xmath26 is twice differentiable at @xmath32 , and @xmath33 ; * if @xmath30 is such that @xmath34 and , in a neighborhood of @xmath35 , @xmath36 for @xmath37 , then @xmath38 and @xmath39 exist and @xmath40 ; * if @xmath30 is such that @xmath34 and , either @xmath38 , or @xmath39 , exists and is different from @xmath41 , then , in a neighborhood of @xmath35 , @xmath36 for @xmath37 . *
if @xmath42 for all @xmath43 in a non - trivial interval @xmath44 , then @xmath45 for every @xmath46 . treating a keplerian system like , we will say that @xmath47 is a @xmath22-bouncing solution if @xmath20 is a @xmath22-bouncing solution of the first differential equation .
we are now ready to state one of the main results of this paper .
[ main ] assume @xmath48 with @xmath49 for a suitable integer @xmath50 .
then , for every integer @xmath51 , there exists an integer @xmath52 , such that for every integer @xmath53 there exists at least one periodic @xmath54-bouncing solution @xmath47 of with period @xmath19 , which makes exactly @xmath18 revolutions around the origin in the period time @xmath19 , i.e. satisfying .
the statement of the theorem requires that the nonlinearity @xmath55 has a _ nonresonant _ asymptotically linear growth at infinity in the following sense .
the constants @xmath56 in are the values of the vertical asymptotes @xmath57 of the @xmath58-th curve of the periodic dancer - fuk spectrum associated to the asymmetric oscillator @xmath59 .
the first equation in presents a _ wall _ at @xmath60
against which the particle bounces .
such a wall can be approximatively modeled as a spring with a very large elasticity constant ( see e.g. @xcite ) . for this reason , it is natural to require that the nonlinearity @xmath55 satisfies an asymptotic behavior at infinity as in .
conditions like have already been introduced treating scalar equations with a singularity , see e.g. @xcite .
we will see also , in theorem [ lltheorem ] , how we can relax condition introducing a landesman - lazer type of condition thus obtaining a similar existence result for nonlinearities _ next to _ resonance .
in particular we will introduce a double resonance condition for nonlinearities satisfying with @xmath61 and @xmath62 .
double landesman - lazer type of condition has been treated in other situations , see e.g. @xcite . moreover , following the proof of theorem [ lltheorem ] , we will obtain corollary [ llcorollary ] which extends to the resonant case a previous result obtained by fonda and the author in @xcite for impact oscillators .
the next section is devoted to the proof of theorem [ main ] . in section [ llsec ]
we will present the results _ next to _ resonance and we will show how to modify the proof of theorem [ main ] in order to prove such a result . then , in section [ seccil ] , we will present how to extend the applications to cylinders .
let us consider the change of coordinate @xmath63 .
system is equivalent to @xmath64 let us define the function @xmath65 as @xmath66 so that the first differential equation in becomes @xmath67 and it is easy to verify that @xmath68 uniformly in @xmath69 $ ] and @xmath70 in a compact set .
in what follows we will assume that @xmath70 varies in a compact set containing zero , but it is not restrictive to assume @xmath70 to be non - negative .
so , fixing @xmath71 , in what follows , we will always assume @xmath72 $ ] .
we are looking for @xmath41-bouncing solutions to such that @xmath73 we define , for a fixed small @xmath74 , for every @xmath75 , the functions @xmath76\times{\mathbb{r}}\times{\mathbb{r}}\to{\mathbb{r}}$ ] @xmath77 and we consider the differential equations @xmath78 where , here , @xmath23 varies in @xmath79 .
let us spend few words to motivate the introduction of the small constant @xmath74 , which can appear unessential : it will be useful to simplify the proof of the validity of the fourth property , which appears in definition [ bsol ] , for the bouncing solution we are going to find ( cf .
@xcite ) .
it is well known that such a differential equation has at least one @xmath4-periodic solution when @xmath80 is large enough and @xmath70 is fixed ( cf .
@xcite ) . in the next section
, we will prove the existence of a common a priori bound .
let us introduce the set of @xmath81 functions which are @xmath4-periodic @xmath82 ) \,:\ , x(0)=x(t ) \
, , x'(0)=x'(t ) \}\,.\ ] ] we are going to look for an open bounded set @xmath83 , containing @xmath41 , such that any periodic solution of belongs to @xmath84 .
then , we will prove in section [ degree ] the existence of periodic solutions of by a limit procedure : we will show that , for every integer @xmath51 and for every integer @xmath85 sufficiently large , there exists a sequence @xmath86 $ ] and a sequence of solutions @xmath87 of , with @xmath88 , converging respectively to @xmath89 and @xmath90 , where @xmath90 is the desired bouncing solution .
the proof of theorem [ main ] will be concluded easily in section [ conclusion ] .
the proof makes use of some phase - plane techniques , so it will be useful to define the following subsets of @xmath91 : @xmath92 and the open balls centered at the origin @xmath93 let @xmath94 be a solution of such that there exist some instants @xmath95 $ ] such that , @xmath96 and @xmath97 .
assume moreover that @xmath98 for every @xmath99 and @xmath100 for every @xmath101 .
let us first consider the interval @xmath102 $ ] .
there exists a positive constant @xmath103 such that the orbit @xmath104 in the phase - plane , in this interval , consists of a branch of the ellipse @xmath105 in particular @xmath106 and @xmath107 , and for every @xmath108 $ ] one has @xmath109 .
a computation shows that , for every @xmath110 , @xmath111 < \frac \pi{\sqrt n}\,.\ ] ] call @xmath112 the open region delimited by the ellipse in .
besides , it is possible to find @xmath113 large enough to have @xmath114 it is useful for our purposes to define the following set @xmath115 in particular we have @xmath116 and @xmath117 we define the following set of periodic functions @xmath118 \ } \,.\ ] ] let us now focus our attention on the second interval @xmath119 $ ] .
it is possible to verify that there exists @xmath120 ( which can be chosen independently of @xmath70 and @xmath80 ) sufficiently large with the following property : if @xmath23 satisfies @xmath121 for every @xmath122 $ ] , then ( cf .
@xcite ) @xmath123 hence , we can conclude that any solution to , such that @xmath124 when @xmath125 , must rotate in the phase plane , and it needs a time @xmath126 , such that @xmath127 \,,\ ] ] to complete a rotation .
so , by , choosing @xmath80 large enough we can find that it performs more than @xmath50 rotations and less than @xmath128 rotations in the phase plane .
in particular it can not perform an integer number of rotations .
moreover , by assumption , if we consider a solution @xmath23 to , satisfying @xmath129 when @xmath125 , introducing polar coordinates @xmath130 , we can find ( enlarging @xmath131 , if necessary ) the following uniform bound for the radial and angular velocity when @xmath23 is positive : @xmath132 @xmath133 thus finding @xmath134 let us now state a lemma which will be useful also in section [ llsec ] .
[ lemmaspirale ] for every @xmath135 there exists @xmath136 and an integer @xmath137 such that every solution @xmath23 to , with @xmath138 satisfying @xmath139 at a certain time @xmath140 $ ] , is such that @xmath141 for every @xmath142 $ ] .
_ it is not restrictive to consider a solution such that @xmath143 and @xmath144 , for every @xmath145 $ ] .
such a solution rotates clockwise in the phase - plane and will vanish for the first time at @xmath146 , thus having @xmath147 such that , by , @xmath148 .
the solution vanishes again at @xmath149 such that @xmath150 .
then , the solution will perform a complete rotation in the interval @xmath119 $ ] thus obtaining @xmath151 . in the time interval
@xmath152 $ ] the solution can not perform more than @xmath128 rotations , so choosing @xmath153 and @xmath154 , as in , we conclude the proof of the lemma . @xmath155 with a similar reasoning , we can prove that if @xmath156 vanishes at a time @xmath157 with @xmath158 then the solution @xmath23 will vanish for the first time at @xmath159 with @xmath160 .
then , the solution will reach a negative minimum at @xmath161 and again will vanish at @xmath162 with @xmath163 . recalling that , in @xmath164 , the orbit of @xmath23
is contained in a certain ellipse of equation we have @xmath165 .
hence , we have immediately , setting @xmath166 @xmath167 such estimates will be useful in section [ llsec ] .
the set @xmath168 provided by the previous lemma is the a priori bound we were looking for .
in fact , suppose to have a @xmath4-periodic solution such that @xmath169 at a certain time @xmath170 $ ] .
if the solution remains outside @xmath171 then it can not perform an integer number of rotations around the origin in the period time @xmath4 .
hence , the solution must enter the set at a certain time @xmath172 , and the previous lemma gives us a contradiction .
summing up , in this section , we have proved the following estimate .
[ finallemma ] there exists a positive integer @xmath173 and an open bounded set @xmath174 such that every @xmath4-periodic solutions to with @xmath175 belong to @xmath176\,\}\,.\ ] ] it is well - known that the existence of periodic solutions of equation is strictly related to the existence of a fixed point of a completely continuous operator @xmath177 ( see e.g. @xcite ) , @xmath178 where @xmath179 is defined in @xmath180 as @xmath181 and @xmath182 does not belong to its spectrum , @xmath183 is the so called nemytzkii operator and @xmath184 is the identity operator . by classical results
( see e.g. @xcite ) one has @xmath185 for every @xmath186 , with @xmath84 given by lemma [ finallemma ]
. using the continuation principle one has , for every @xmath186 , that there exists a continuum @xmath187 in @xmath188\times\omega$ ] , connecting @xmath189 to @xmath190 , whose elements @xmath191 are such that @xmath192 is a solution of @xmath193 ( see @xcite for a similar approach ) . the function @xmath194\times\omega \to { \mathbb{r}}$ ] @xmath195
is well defined and continuous , being @xmath196 , by lemma [ finallemma ] . in particular , one has @xmath197 and @xmath198 where @xmath199 is the constant provided by lemma [ lemmaspirale ] .
hence , for every integer @xmath51 , there exists @xmath200 with the following property : for every @xmath53 and for every @xmath186 , there exists @xmath201 with @xmath202 .
fix now @xmath18 and @xmath53 and consider the sequences @xmath203 and @xmath204 .
let us simply denote them by @xmath205 and @xmath206 .
both the sequences are contained in a compact set respectively of @xmath79 and @xmath207 , so there exist , up to subsequences , @xmath208 and @xmath209 such that @xmath210 and @xmath211 uniformly ( the estimate in gives us that @xmath212 is positive ) . moreover , by continuity of @xmath213 , we have @xmath214
. we have to prove that @xmath209 is a bouncing solution of the differential equation , where @xmath215 .
the proof follows the same procedure of the one in @xcite considering the sequence of approximating differential equations @xmath216 for briefness we refer to that paper for the proof of this part .
we have found in the previous section , for every integer @xmath18 and for every integer @xmath85 sufficiently large , a solution @xmath23 of and @xmath217 $ ] such that @xmath218 .
so , defining for a certain @xmath219 , @xmath220 we have that @xmath221 is a bouncing periodic solution of and using the change of coordinate @xmath222 we find the bouncing solution of satisfying .
the proof of theorem [ main ] is thus completed .
in this section we will see how we can relax the hypotheses of theorem [ main ] in order to treat the situation of a nonlinearity which has an asymptotically linear growth next to resonance .
we are going to provide a landesman - lazer type of condition for the case when the nonlinearity @xmath55 satisfies with @xmath223 and @xmath62 , where @xmath224 is the vertical asymptote of the @xmath58-th curve of the dancer - fuk spectrum .
such a situation has been often called as a _ double resonance _ ( see e.g. @xcite ) .
one - side _ landesman - lazer condition for the scalar differential equation with singularity has been provided by fonda and garrione in @xcite .
this section has been inspired by such a paper and some steps of the proof of lemma [ lemmall ] will appear similar .
the main novelty occurs in the estimate in and its proof , permitting us to treat a double resonance situation .
in particular , the validity of , permits us to obtain a landesman - lazer condition involving the function @xmath225\ ] ] extended by periodicity to the whole real line , while , in @xcite , the landesman - lazer condition is weaker , being related to a function of the type @xmath226 \\[2 mm ] 0 & t\in \left[\frac{t}{j } , t\right ] \end{cases}\ ] ] extended by periodicity . we will prove in this section the following result .
[ lltheorem ] assume that there exists a constant @xmath227 such that , for @xmath228 , @xmath229 for every @xmath230 $ ] and every @xmath231 .
moreover , for every @xmath232 $ ] , @xmath233 and @xmath234 then , for every integer @xmath51 , there exists an integer @xmath52 , such that for every integer @xmath53 there exists at least one periodic @xmath54-bouncing solution @xmath47 of with period @xmath19 , which makes exactly @xmath18 revolutions around the origin in the period time @xmath19 , i.e. satisfying
. the case @xmath235 will be briefly treated at the end of this section .
first of all we need to introduce a sequence of approximating equations with non - resonant nonlinearities .
then , we will look for a common a priori bound .
it is possible to find , when @xmath80 is chosen large enough , a constant @xmath236 such that @xmath237 in this way , we have @xmath238 we define the function @xmath239 , similarly as in , @xmath240 where @xmath241 we will consider the differential equation @xmath242 notice that @xmath243 thus giving us , by the previous computation in , that the nonlinearities @xmath239 are nonresonant .
so , if we find a common a priori bound @xmath84 as in lemma [ finallemma ] , uniform in @xmath70 and @xmath80 , for every @xmath4-periodic solution of , then holds , thus permitting us to end the proof of the theorem as in section [ degree ] and [ conclusion ] .
the landesman - lazer type of conditions introduced in and are needed in order to find the common a priori bound for every @xmath80 sufficiently large .
we can assume without loss of generality that the constant @xmath227 in the statement of the theorem guarantees also that , when @xmath244 , for every @xmath70 , @xmath43 and @xmath80 , @xmath245 the reader will notice that the next result comes out free by the proof of the previous theorem ( setting @xmath246 everywhere ) .
such a corollary extends a previous result provided by fonda and the author in @xcite .
[ llcorollary ] assume that there exists a constant @xmath227 such that @xmath247 for every @xmath230 $ ] and every @xmath248 .
moreover , for every @xmath232 $ ] , @xmath249 and @xmath250 then , there exists a @xmath41-bouncing solution for equation @xmath251 . repeating the reasoning explained in section [ secnonres ] , providing the estimate in , we can find , for every @xmath252 , a value @xmath253 such that every solution @xmath23 of
, satisfying @xmath254 when @xmath125 , must rotate in the phase plane spending a time @xmath255 with @xmath256 when @xmath257 is chosen sufficiently small , one has that a @xmath4-periodic solution @xmath23 of , such that @xmath254 when @xmath125 , must perform exactly @xmath50 or @xmath128 rotations around the origin .
the previous reasoning holds uniformly for every @xmath80 and every @xmath70 , so we can fix such a suitable @xmath257 and find the constant @xmath258 .
we underline that lemma [ lemmaspirale ] still holds under the hypotheses of theorem [ lltheorem ] , too .
we underline that remains valid .
the needed a priori bound is given by the following lemma .
[ lemmall ] there exist @xmath259 ( given by lemma [ lemmaspirale ] ) and @xmath260 , as in , such that , for every @xmath175 and every @xmath72 $ ] , any @xmath4-periodic solution @xmath23 of is such that @xmath261 when @xmath262 . in particular , we immediately have @xmath263 . _ proof .
_ suppose by contradiction that there exist an increasing sequence @xmath264 with @xmath265 and @xmath266 , an increasing sequence @xmath267 of integers @xmath268 , a sequence @xmath269 $ ] , a sequence @xmath270 of solutions to @xmath271 and a sequence of times @xmath272 $ ] such that @xmath273 and @xmath274 .
being @xmath275 , thanks to lemma [ lemmaspirale ] , all the solutions can not enter @xmath171 , so that they must perform exactly @xmath50 or @xmath128 rotations around the origin .
we define the sequence of functions @xmath276 which are solutions of @xmath277 by , we have @xmath278 \text { and } \frac 1{c_0 } < \|v_m'\|_\infty < c_0 \,,\ ] ] for a suitable @xmath279
. we have immediately that @xmath280 is bounded in @xmath281 so that , up to subsequences we have @xmath282 weakly in @xmath283 and uniformly , moreover we can assume that @xmath284 . in particular @xmath285 , being @xmath286 , and it is non - negative and @xmath4-periodic .
we assume that , up to subsequence , all the solutions make exactly @xmath128 rotations .
we will discuss the other situation later on .
we are going now to prove that @xmath287 solves @xmath288 for almost every @xmath43 .
call @xmath289 which is an at most countable union of open interval .
consider a function @xmath290 with compact support @xmath291 .
multiplying by @xmath290 and integrating in @xmath292 we obtain @xmath293 by compactness we have @xmath294 , for a suitable @xmath295 , thus giving us that @xmath296 uniformly for @xmath297 .
assuming , up to subsequence , @xmath298 for every @xmath297 we can find , for every @xmath299 an index @xmath300 such that @xmath301 hence , the subsequence @xmath302 is bounded in @xmath303 , so up to a subsequence it converges weakly to a certain function @xmath304 such that @xmath305 for almost every @xmath297 almost everywhere .
hence , passing to the limit in , we obtain @xmath306 it is possible to extend the function @xmath307 to the whole set @xmath308 , so that @xmath309 thus giving us that @xmath287 is a _ weak _ solution of @xmath310 in @xmath308 .
in particular @xmath311 and @xmath312 .
we must show that @xmath313 for almost every @xmath314 .
we recall that the functions @xmath315 perform in the phase plane exactly @xmath128 rotations around the origin , so there exist @xmath316 such that , for every @xmath317 , @xmath318 @xmath319 up to subsequences , we can assume that @xmath320 and @xmath321 such that @xmath322 being @xmath323 , then @xmath324 .
moreover , using the estimate in , from @xmath325 , we have @xmath326 . from @xmath327
we have @xmath328 , where we denote @xmath329 .
we now consider an interval @xmath330 $ ] with @xmath331 and @xmath332 in @xmath333 .
writing in polar coordinates @xmath334 we obtain the expression of the angular velocity of the solution in the phase plane @xmath335 thus obtaining @xmath336 so that @xmath337 the previous reasoning can be done for every interval @xmath338 $ ] , so the only reasonable conclusion is that @xmath339 .
hence , if the solution @xmath287 becomes positive , then it is forced to remain positive for a time equal to @xmath340 .
passing to modified polar coordinates @xmath341 we obtain integrating @xmath342 on @xmath330 $ ] @xmath343 thus giving us @xmath344 for almost every @xmath345 $ ] .
in particular , @xmath308 is the union of the intervals @xmath346 , where we have @xmath347 with @xmath348 $ ] and at least one of them is equal to @xmath349 , being @xmath350 .
we now prove that @xmath351 the functions @xmath352 solve equation , which we rewrite in the simpler form @xmath353 where , for every @xmath354 , @xmath355 .
we show that , if @xmath287 is positive in @xmath356\subset[0,t]$ ] then @xmath352 @xmath81-converges to @xmath287 .
we have already seen that @xmath280 is bounded in @xmath81 , and by as an immediate consequence we get @xmath357 for every @xmath358 $ ] .
so , being @xmath352 bounded in @xmath359 in such a interval , by the ascoli - arzel ` a theorem , we have that @xmath352 @xmath81-converges to @xmath287 in @xmath356 $ ] .
the @xmath81-convergence and the estimate in are the ingredients we need to prove that the solution @xmath287 has only isolated zeros @xmath360 .
we start proving that if the left derivative @xmath361 for a certain index @xmath362 , then @xmath363 for @xmath354 large enough .
for every @xmath364 , we can find @xmath365 sufficiently small to have @xmath366 for every @xmath367 .
being @xmath352 @xmath81-convergent to @xmath287 in @xmath368 , for @xmath354 large enough , @xmath369 for every @xmath367 .
being @xmath370 in this interval , we find @xmath371 choosing @xmath372 sufficiently small .
we prove now that if @xmath361 for a certain index @xmath362 , then @xmath360 is an isolated zero of @xmath287 .
suppose by contradiction that there exists @xmath373 , with @xmath374 as in , such that @xmath375 .
for every @xmath354 large enough we have @xmath376 and by the previous computation @xmath377 .
the property that @xmath378 follows directly by the fact that the nonlinearities @xmath239 do not depend by @xmath43 when @xmath379 , leads us to treat a landesman - lazer condition involving function @xmath380 as in rather than in . here lies one of the main differences between our result and the one obtained by fonda and garrione in @xcite .
being @xmath381 when @xmath352 is positive , we can show that if @xmath382 then @xmath383 .
by construction @xmath384 for a certain @xmath385 , so that we obtain @xmath386 for every @xmath354 large enough , thus contradicting @xmath282 .
being @xmath287 as in , and @xmath387 for at least one value @xmath317 , for such index @xmath388 holds .
the previous reasoning gives us that @xmath389 . iterating the procedure we can prove that @xmath390 for every index @xmath317 .
we prove now that for every @xmath317 , the left and right derivatives satisfy @xmath391 , thus we can conclude that @xmath387 for every @xmath317 .
suppose by contradiction that there exists @xmath317 such that @xmath392 without loss of generality we suppose this value to be positive .
the other case follows similarly .
so , assume @xmath393 arguing as above , for every @xmath364 , we can find @xmath365 sufficiently small to have @xmath394 for every @xmath367 .
being @xmath352 @xmath81-convergent to @xmath287 in @xmath368 and in @xmath395 , for @xmath354 large enough , @xmath396 for every @xmath367 .
being @xmath370 , we find @xmath397 ^ 2 + 4\epsilon_0 d \geq v_m'^2(\beta_r^m)= v_m'^2(\alpha_{r+1}^m ) \geq [ v'(\xi_r^+ ) - 2\epsilon_0]^2 - 4\epsilon_0 d\,.\ ] ] hence @xmath398 thus giving us a contradiction for @xmath372 sufficiently small .
we have proved that @xmath287 , in every interval @xmath330=[\xi_r,\xi_{r+1}]$ ] , satisfies @xmath399 so , @xmath287 is a solution of the following dirichlet problem : @xmath400 let us consider the orthonormal basis @xmath401 of @xmath402 made of the eigenfunctions solving the dirichlet problem @xmath403 where @xmath404 is the @xmath85-th eigenvalue .
denoting by @xmath405 and @xmath406 , respectively , the scalar product and the norm in @xmath402 , we can write the fourier series of all the functions @xmath407 and split it as follows : x_m & = & _ k=1^ x_m , _ k _ k + & = & _ x_m^0 + _ x_m^ , with the following property @xmath408 moreover , one has @xmath409 with @xmath410 being @xmath411 , we have @xmath412 uniformly in @xmath330 $ ] .
moreover @xmath413 , for @xmath354 sufficiently large .
multiplying equation by @xmath414 and integrating in the interval @xmath330 $ ] we obtain @xmath415 { \displaystyle}\hspace{10mm}= - \int_\alpha^\beta x_m^0(t ) ( v_m^0)''(t ) \,dt = \int_\alpha^\beta \mu_{n+1 } x_m^0(t ) v_m^0(t ) \,dt \\ [ 3 mm ] { \displaystyle}\hspace{20mm}= \int_\alpha^\beta \mu_{n+1 } x_m(t ) v_m^0(t ) \,dt \ , . \end{array}\ ] ] defining @xmath416 we have @xmath417 and applying fatou s lemma @xmath418 it is easy to see that for every @xmath419 it is possible to find @xmath420 such that @xmath421 for every @xmath422 . hence , being @xmath412 and @xmath423 , we have @xmath424 v(t ) \,dt \geq 0\,.\ ] ] the previous estimate can be obtained for every interval @xmath346 , thus contradicting setting @xmath425 .
we have so finished to prove the case in which the sequence @xmath270 consists of solutions performing @xmath128 rotations in the phase - plane around the origin for an infinite number of index @xmath354 .
we now treat the case in which the solutions perform @xmath50 rotations around the origin . in this case
we have to prove that the limit function @xmath287 solves @xmath426 for almost every @xmath43 . with the same procedure we can show that there exists a @xmath427-function @xmath428 , satisfying @xmath429 almost everywhere , such that @xmath287 is a weak solution of @xmath430 .
then , with a similar procedure , it is possible to introduce some instants @xmath431 , @xmath432 , when @xmath352 vanishes , converging to some values @xmath360 ( with @xmath433 ) . unfortunately .
we can not conclude immediately that @xmath434 holds .
in fact , using , we can verify only that @xmath435 .
nevertheless , we have that @xmath352 @xmath81-converges to @xmath287 when @xmath287 is positive . moreover ,
following the reasoning which gave us the estimates in , we can prove that for at least one index @xmath433 the left - derivative satisfies @xmath436 ( e.g. the index in which @xmath287 attains the maximum ) .
arguing as above , we can prove that whenever @xmath436 for a certain index @xmath433 then @xmath360 is an isolated zero and @xmath437 .
iterating this reasoning , we have that all the @xmath360 are isolated zeros of @xmath287 thus obtaining the needed estimate @xmath438 .
now , with a similar procedure , we can prove that @xmath439 , thus giving us that @xmath440 in the interval @xmath346 .
also in this case , in the same way , it is possible to conclude that @xmath387 for every @xmath433 .
then , we can consider an interval @xmath330 $ ] with @xmath441 in @xmath333 , and with a similar reasoning we can obtain a _
estimate similar to the one obtained , thus gaining a contradiction with assumption .
the proposition is thus proved .
let us spend few words about the possibility of extending theorem [ lltheorem ] to the case @xmath235 , where @xmath442 and @xmath443 .
there is a wide literature ( cf .
@xcite ) treating nonlinearities _ lying _ under the first curve of the dancer - fuk spectrum .
the landesman - lazer condition in this case reduces to a sign condition on the nonlinearity @xmath55 .
unfortunately , it is not possible to obtain a proof with the same procedure , cf .
the estimate in . for briefness we do not enter in such details in this paper . however , let us state the following weaker result for a nonlinearity with a _
one - side _ resonance condition , the proof of which works similarly to the one of theorem [ lltheorem ] .
[ lltheoremn0 ] assume that there exists a constant @xmath444 such that @xmath445 uniformly for every @xmath230 $ ] and that there exists a constant @xmath227 such that , for @xmath228 , @xmath446 for every @xmath230 $ ] and every @xmath231 .
moreover , for every @xmath232 $ ] , @xmath447 then , for every integer @xmath51 , there exists an integer @xmath52 , such that for every integer @xmath53 there exists at least one periodic @xmath54-bouncing solution @xmath47 of with period @xmath19 , which makes exactly @xmath18 revolutions around the origin in the period time @xmath19 , i.e. satisfying .
in this section we briefly explain how the previous results could be applied to a class of systems defined in @xmath448 , modeling a particle hitting a cylinder @xmath9 . the case @xmath10 and @xmath11 models bounces on a proper cylinder . for briefness we will present the result for _ non - resonant _ nonlinearities .
we consider the differential equations @xmath449 where @xmath450 and @xmath451 .
we assume for simplicity all the functions to be continuous .
we suppose that @xmath452 satisfies the assumptions of theorem [ main ] .
the function @xmath453 satisfies @xmath454 with @xmath455 uniformly in @xmath43 and * y*. assume that the second equation in can be viewed , in every components , as @xmath456 where @xmath457 uniformly in all the other variables .
we can assume , as an example of application , under the previous assumptions , for every integer @xmath466 , there exists an integer @xmath467 , such that for every integer @xmath468 there exists at least one periodic solution @xmath469 of , such that * x * , can be parametrized in polar coordinates @xmath47 and @xmath20 is a @xmath54-bouncing solution .
such solutions satisfies the following periodicity conditions : @xmath470 the proof of such a result can be obtained by glueing together the results contained in this paper ( for the @xmath471 coordinate ) and classical results ( for the @xmath472 coordinate ) .
the key tool is the fact that the equations are _ weakly _ coupled . , _ periodic motions of an impact oscillator _ , j. sound vibration * 209 * ( 1998 ) , 4360 . , _ chaos in forced impact systems _ , discrete contin .
dyn . syst .
* 6 * ( 2013 ) , 861890 .
, _ periodic motions in impact oscillators with perfectly elastic bounces _ , nonlinearity * 15 * ( 2002 ) , 12811297 . , _ on the approximation of the elastic bounce problem on riemannian manifolds _ , j. differential equations * 47 * ( 1983 ) , 227245 . , _ uniqueness of the one - dimensional bounce problem as a generic property in @xmath473 ; { \mathbb r})$ ] _ , boll .
ital . a ( 6 ) * 1 * ( 1982 ) , 8791 . , _
@xmath4-periodic solutions for some second order differential equations with singularities _ , proc .
edinb . * 120a * ( 1992 ) , 231243 . , _ on the periodic boundary value problem for
forced duffing equations with jumping nonlinearity _
, nonlinear anal . *
19 * ( 1986 ) , 643650 . , _ landesman - lazer conditions for periodic boundary value problems with asymmetric nonlinearities _ , j. differential equations * 116 * ( 1995 ) , 405418 . , _ periodic solutions of nonlinear differential equations with double resonance _ , ann .
pura appl .
* 157 * ( 1990 ) , 99116 . , _ periodic solutions of second order differential equations with superlinear asymmetric nonlinearities _ , arch . math .
* 60 * ( 1993 ) , 266276 . , _ periodic solutions of a second order differential equation with one - sided growth restriction on the restoring term _ , arch
* 51 * ( 1988 ) , 151163 . , _
double resonance with landesman
lazer conditions for planar systems of ordinary differential equations _
, j. differential equations * 250 * ( 2011 ) , 10521082 . , _ a landesman - lazer type of condition for asymptotically linear secondo order equations with a singularity _ , proc .
* 142 * ( 2012 ) , 12631277 . , _ planar differential systems at resonance _ ,
differential equations * 11 * ( 2006 ) , 11111133 . , _ a general method for the existence of periodic solutions of differential equations in the plane _ ,
j. differential equations * 252 * ( 2012 ) , 13691391 . , _ periodic bouncing solutions for nonlinear impact oscillators _ , adv .
nonlinear stud .
* 13 * ( 2013 ) , 179189 . , _ periodic orbits of radially symmetric keplerian - like systems : a topological degree approach _ , j. differential equations * 244 * ( 2008 ) , 32353264 . , _ radially symmetric systems with a singularity and asymptotically linear growth _ , nonlinear analysis * 74 * ( 2011 ) , 24852496 . , _ periodic solutions of radially symmetric perturbations of newtonian systems _ , proc .
soc * 140 * ( 2012 ) , 13311341 .
, _ periodic solutions of singular radially symmetric systems with superlinear growth _
, ann . mat .
pura appl .
* 191 * ( 2012 ) , 181204 . ,
_ a necessary and sufficient condition of nonresonance for a semilinear neumann problem _ , proc .
* 114 * ( 1992 ) , 433442 . , _ nonresonance conditions on the potential with respect to the fuk spectrum for the periodic boundary value problem _ , rocky mountain j. math .
* 25 * ( 1995 ) , 13051340 . , _ periodic solutions of second order differential equations with an obstacle _ , nonlinearity * 19 * ( 2006 ) , 11651183 . , _ periodic motions in the fermi - ulam model _ , proc .
* 138a * ( 2008 ) , 12351259 . ,
_ chaotic regular and unbounded behaviour in the elastic impact oscillator _ , physica d * 82 * ( 1995 ) , 117135 . , _
periodic bouncing for a forced linear spring with obstacle _ , differential integral equations * 5 * ( 1992 ) , 165172 .
, _ periodic solutions of some forced linard differential equations at resonance _
, arch . math .
* 41 * ( 1983 ) , 337351 . , _ dynamics of a forced oscillator having an obstacle _ , in variational and topological methods in the study of nonlinear phenomena ( pisa , 2000 ) , 7587 , progr . in nonlinear differential equations
49 , birkh " auser , boston , 2002 . , _ linear motions in a periodically forced kepler problem _ , portugal .
* 68 * ( 2011 ) , 149176 . , _ large amplitude periodic bouncing for impact oscillators with damping _ , proc .
* 133 * ( 2005 ) , 17971804 . , _ periodic motions of linear impact oscillators via the successor map _ , siam j. math . anal .
* 36 * ( 2005 ) , 17071725 .
, _ periodic solutions and chaotic dynamics in forced impact oscillators _ , siam j. appl .
* 2013 * ( 12 ) , 383414 . | we investigate the existence of periodic trajectories of a particle , subject to a central force , which can hit a sphere , or a cylinder
. we will provide also a landesman - lazer type of condition in the case of a nonlinearity satisfying a _ double _ resonance condition .
afterwards , we will show how such a result can be adapted to obtain a new result for the impact oscillator at _
double _ resonance . |
the study of quantum field theory ( qft ) in @xmath0-dimensions provides valuable insights into many difficult conceptual problems of qft s , as well as giving us situations where appropriate use of infinite dimensional symmetries , like conformal invariance and @xmath1 symmetry , considerably simplifies the field theoretic computation .
one might ask if similar insights could also be obtained by studying qft s in @xmath0-dimensional noncommutative spacetime .
indeed there is reason for some optimism regarding this query , as it is possible to define notions of conformal invariance , kac - moody and virasoro symmetries @xcite . in this article
we will address the question of quantum integrability , and investigate the noncommutative analog of the quantum sinh - gordon model .
we will argue that it is integrable , in the sense that there is no particle production , and calculate the exact two - particle scattering matrix . in section 2
, we will summarize the results on twisted qft s based on our earlier work @xcite , and then review scattering theory for noncommutative qft s @xcite . in section 3 , we will recall some essential features of the quantum sinh - gordon model on commutative spacetime , and then go on to construct the two - particle @xmath2-matrix @xmath3 for its noncommutative counterpart . a discussion of the properties of @xmath3 will be provided in section 4 .
two dimensional noncommutative spacetime is generated by operators @xmath4 satisfying the commutation relations @xmath5 = i \theta_{\mu\nu }
\equiv i \theta \epsilon_{\mu\nu } , \quad \mu,\nu = 0,1 \ , \label{ncrelation}\ ] ] where @xmath6 are hermitian operators , and @xmath7 is a real constant . using the moyal map , we can map operators built out of ( the products of ) the @xmath4 s to functions on minkowski space @xmath8 , but with a modified rule for multiplication . if @xmath9 and @xmath10 are two operators that are mapped to functions @xmath11 and @xmath12 respectively , then their _ star - product _ consistent with ( [ ncrelation ] ) is @xmath13 for example , for @xmath14 , @xmath15 notice that the commutation relation ( [ ncrelation ] ) is unchanged under translations @xmath16 and identity - connected lorentz transformations @xmath17 .
thus poincar transformations , with their usual action on coordinates , belong to the automorphism group of noncommutative @xmath18 . our interest in this noncommutative space is to understand the notion of integrability for quantum theories . as long as the hamiltonian is ( formally ) hermitian , the quantum theory on such a space is manifestly unitary , as emphasized in the case of field theories by @xcite , and demonstrated for single particle quantum mechanics by @xcite .
we will therefore proceed to write free quantum fields , which we will use to build interactions .
a free quantum scalar field @xmath19 can be expanded as @xmath20 while it is possible to quantize the field by imposing the standard ( or canonical ) commutation relations between @xmath21 s and @xmath22 s , more general commutation relations are possible as well @xcite .
these _ twisted _ commutation relations are of the form : @xmath23 where @xmath24 is any lorentz - invariant function of two - momenta @xmath25 and @xmath26 . in @xcite
, we argued that in spacetime dimension greater than 2 , compatibility with quantum statistics of identical particles requires @xmath24 to be of the form @xmath27 we will work henceforth with this choice of @xmath24 , and argue in section 3 why this leads to integrability of the @xmath2-matrix for the noncommutative sinh - gordon model .
even though the commutation relations ( [ twist1][twist3 ] ) are twisted , the operators @xmath28 act on the usual ( bosonic ) fock space . to see this ,
let @xmath29 be the ordinary or untwisted creation and annihilation operators : @xmath30 = 0 , \quad [ c_p , c^\dagger_q ] = 2p_0 \delta(\vec{p } - \vec{q } ) .
\label{usualcomm}\ ] ] there is a simple relation between the @xmath31 s and the @xmath32 s ( for @xmath33 ) . to this end , consider the fock space momentum operator @xmath34 associated with @xmath19 : @xmath35 & = & -i\partial_\mu \phi(x)\end{aligned}\ ] ] using @xmath34 , we can realize the twisted creation - annihilation operators in terms of the untwisted ones : @xmath36 it is easy to check that ( [ atoc1 ] , [ atoc2 ] ) reproduce the twisted commutation relations ( [ twist1][twist3 ] ) .
thus the @xmath32 s act on the same fock space as that of the untwisted creation and annihilation operators .
the map ( [ atoc1],[atoc2 ] ) is a `` dressing transformation '' ( first discussed in @xcite ) , and the commutation relations for the @xmath32 s and @xmath37 s that follow from it are simple examples of the algebra discussed in@xcite
. a twisted @xmath38-particle state is @xmath39 this is related very simply to the ordinary @xmath38-particle state . using ( [ atoc2 ] ) , @xmath40 in @xmath0 dimensions , it is often convenient to work in light - cone coordinates , with the two - momentum characterized by _
@xmath41 : @xmath42 . the twisted commutation relation ( [ twist3 ] )
then becomes @xmath43 finally , the quantum field @xmath19 even though free , is not local : the commutator @xmath44 $ ] is non - zero for @xmath45 and @xmath46 space - like separated @xcite .
consider a theory with interaction hamiltonian @xmath47 the scattering operator for this theory is @xmath48 the first non - trivial term in the perturbative expansion of the above is @xmath49 using the mode expansion for @xmath19 , let us look at a typical term in the above , which is of the form @xmath50 . using ( [ atoc2 ] ) and ( [ planewavestar ] ) , @xmath51 where we have integrated by parts and discarded surface terms to obtain the last expression .
but ( [ cs ] ) is just the same as the corresponding term from the commutative scattering theory .
hence to order @xmath52 , the noncommutative scattering operator is the same as in the commutative theory . more generally , this is true to any order in perturbation theory ( see @xcite for details of the proof ) . in particular
, this means that @xmath53 the scattering operator on noncommutative space is the same as that for the commutative counterpart . in particular ,
if a commutative theory is renormalizable , so is its noncommutative counterpart , because the number of counterterms is the same .
as far as scattering is concerned , the main difference between a commutative theory and its noncommutative counterpart is in the nature of asymptotic states .
but since we know the explicit map ( [ statemap ] ) between these two kinds of asymptotic states , the matrix elements of @xmath3 can be calculated in terms of those of @xmath54 .
the result ( [ soperator ] ) is true for theories without interacting non - abelian gauge fields only @xcite .
for example , in noncommutative @xmath55 there are effects that violate lorentz invariance .
however , this caveat is not applicable here , as we will only consider theories where the interaction hamiltonian is made up of matter fields only .
one of the the simplest non - trivial integrable model in @xmath0-dimensional commutative spacetime is the sinh - gordon model .
this is the theory with interaction hamiltonian of the form @xmath56 where the double dots above stand for normal - ordering .
for this theory , there is no particle production : the amplitude @xmath57 for producing @xmath38 outgoing particles by colliding @xmath58 incoming particles in zero if @xmath59 .
in addition , the `` elastic '' amplitude @xmath60 factorizes into products of two - particle scattering amplitudes @xmath61 .
the exact expression of the two - particle @xmath2-matrix on commutative spacetime has been given in @xcite : @xmath62}{\tanh \left[\frac{1}{2 } \left(\eta + i \frac{\pi}{2}b(g)\right ) \right ] } , \quad { \rm where } \quad b(g ) = \frac{2g^2}{8\pi + g^2}\ ] ] and @xmath41 the relative rapidity . a very nice argument motivating quantum integrability for the commutative case has been given by dorey in @xcite .
we will use this argument , adapting it appropriately to the noncommutative case . consider the theory based on the free field @xmath19 as in ( [ tfield ] ) , with the free hamiltonian @xmath63 , and an interaction @xmath64 , where by @xmath65 we mean @xmath66 .
there are two diagrams ( figure [ fig:2to4 - 1 ] ) that contribute to the @xmath67 scattering amplitude at tree level .
the amplitude for this process is @xmath68 where we have used ( [ statemap ] , [ soperator ] ) , and the fact that @xmath69 .
let us add an extra interaction of the form @xmath70 .
we see that the @xmath67 process receives an additional contribution from a third diagram ( figure [ fig:2to4 - 2 ] ) , which of the form @xmath71 by choosing @xmath72 , we can make the total tree - level amplitude @xmath73 for the @xmath67 process to vanish . for the new interaction hamiltonian @xmath74 , the amplitude for the @xmath75 process is now a non - zero constant , which can be made to vanish by judiciously choosing an extra interaction piece of the form @xmath76 .
continuing in this manner , one finds that the theory with the interaction hamiltonian of the form @xmath77 has no particle production at tree - level : all processes of the form @xmath78 @xmath79 are forbidden ( we have moved the mass term @xmath80 from the free hamiltonian to the interaction hamiltonian , so that it can be presented in the convenient form ( [ ncsg ] ) . ) a crucial ingredient in the argument above is the specific choice ( [ gform ] ) of the twist function @xmath24 for the commutation relations of the free field creation and annihilation operators .
had we chosen it to be of some other form , it is easy to see that we would lose the _ no particle production _ condition , and hence quantum integrability . in particular , choosing the twist function to be identity ( i.e. using conventional commutation relations ( [ usualcomm ] ) for the free field creation / annihilation operators ) leads to particle production at tree - level itself , as shown explicitly by @xcite ( however , see also @xcite for a discussion of the absence of tree - level particle production in the noncommutative _ sine_-gordon model ) .
our argument also extends to higher loops . for a scalar field theory ( with , say , polynomial interactions ) in commutative @xmath0 dimensions , the only source of ultraviolet divergence in perturbation theory is from single closed loops ( see for example @xcite ) .
these can be absorbed by renormalizing the mass of the particle , or equivalently by working with the normal - ordering the interaction hamiltonian to start with , as we have done . as we argued earlier , although the scattering operator for the noncommutative theory and its commutative counterpart is the same , the asymptotic states are different . using the map ( [ statemap ] ) , we find the two - particle amplitude for the noncommutative case to be @xmath81}{\tanh \left[\frac{1}{2 } \left(\eta + i \frac{\pi}{2}b(g)\right ) \right ] } e^{-i\theta m^2 \sinh \eta}\ ] ] where @xmath82 is the ( physical ) mass of the sinh - gordon particle .
it is obvious that @xmath83 satisfies the following conditions : the fact the noncommutative scattering matrix for the sinh - gordon model differs from commutative one only by a phase may seem surprising at first .
it was pointed out by mitra @xcite that overall phases of the form @xmath90 are allowed , over and above the form of the @xmath2-matrix dictated by dynamics . for local fields ,
the @xmath91 are required to vanish . however , since our field is non - local , we have no such restriction .
we find , in fact , that @xmath92 . | using twisted commutation relations we show that the quantum sinh - gordon model on noncommutative space is integrable , and compute the exact two - particle scattering matrix .
the model possesses a strong - weak duality , just like its commutative counterpart .
iisc / chep/10/07 + * the quantum sinh - gordon model in noncommutative ( 1 + 1 ) dimensions * sachindeo vaidya + |
softly broken gauge symmetries are frequently an important property in quantum field theory ( qft ) .
one example is supersymmetry , which must be ( most likely softly ) broken in order to address the phenomenological applications and eventually experimental tests @xcite .
another interesting application of the softly symmetry breaking is the effective qft approach to the propagating torsion @xcite .
the completely antisymmetric component of torsion can be described by the dual axial vector coupled to fermions through the axial vector current .
the presence of the symmetry breaking mass of the axial vector is required for the consistency of the effective theory in the low - energy sector .
indeed , the massive couterterm shows up at 1-loop level . in many cases ,
one is interested not only in the classical aspects of the theory , but also in the derivation of quantum corrections .
the subject of the present paper is the calculation of 1-loop effective action for the softly broken gauge theory of propagating torsion in curved space - time . here , the kinetic term and the interactions terms in the classical action are gauge invariant while the massive terms are not .
consequently , the standard methods for evaluating the effective action face serious technical difficulties . as a strategy
, we shall apply the st@xmath0ckelberg procedure @xcite , that is , we are going to restore the gauge symmetry by introducing an extra field or a set of fields .
more details and applications of the method to models with softly broken gauge symmetry in curved spacetime can be found in ref .
@xcite .
we are going to show that our approach means much simpler and more efficient calculation of quantum corrections .
the difference is especially explicit for the massive torsion - fermion system which was originally elaborated in ref .
the present method provides an independent verification of our previous result in ref .
@xcite and also enables one to perform the calculations in an arbitrary curved space - time , something that was impossible in the framework used in ref .
torsion @xmath1 is an independent ( along with the metric ) quantity describing the spacetime manifold .
it is defined by the relation ( see , e.g. , refs .
@xcite for introduction ) @xmath2 it proves useful to divide torsion into three irreducible components @xmath3 as already known in literature . the interaction with the dirac fermion
is described in a quantum consistent way by the action for the theory of effective fermion - torsion system @xcite ( see also refs .
@xcite ) , s_tf = d^4x \ { - 14s_^2 - 12m^2 s^2_+ i|^ ( _ + i ^5s _ ) + m|}. [ t5 ] here @xmath4 , @xmath5 is the torsion mass , we consider only one non - vanishing component of torsion , @xmath6 , and @xmath7 is the covariant derivative without torsion . to calculate the 1-loop effective action for this model , one has to apply the generalized method of schwinger - dewitt @xcite in the transverse vector space , as was done in ref .
@xcite . following the approach discussed in ref .
@xcite , one can apply the st@xmath0ckelberg procedure by introducing a new scalar field , @xmath8 , and restoring the gauge symmetry in the following way : s_tf^ & = & d^4x \ { - 14s_^2 + 12m^2(s_- ) ^2 + & + & i|^ ( _ + i _ 1 ^ 5s _ ) + m | ( ) } , [ t555 ] the gauge symmetry must be supplemented by @xmath9 .
the original theory ( [ t5 ] ) is restored when we use the gauge fixing condition @xmath10 .
in order to obtain the one - loop divergences for the original theory ( [ t5 ] ) , one has to put @xmath11 in the general expression for the divergences of theory ( [ t555 ] ) , which can be computed by the standard schwinger - dewitt method .
then the final result reduces to ^(1)_div & = & - d^nx\ { 4 ^ 2 m^2 s^s_- 13 ^2 s_^2 + 4i^2|^_^+ .
2i^2|^_+ ( - 4 ^ 2 m ) |+ |r + ( |)^2 } , + [ r1 ] where @xmath12 and @xmath13 .
it is worth mentioning that the above result is more general than the result of ref .
indeed , it is valid in curved spacetime , where a new non - minimal coupling with curvature shows up .
of course this term is relevant for dynamics of dirac particles in the curved background , but the theory contains the @xmath14-term which has non - trivial consequences .
in fact , at 2-loop level , this term is responsible for appearance of the feynman diagrams drawn in fig . 1
. detailed calculation of these diagrams reveals the appearance of the @xmath16-type counterterm , which introduces longitudinal degrees of freedom breaking unitarity .
this undesireble contribution can not be compensated by another 2-loop diagrams , unless some artificial fine - tunning between different coupling constants takes place .
even if theory ( [ t5 ] ) is not consistent at the quantum level , it can pehaps mimic some fundamental theory , as an effective theory . in this sense
, it would be interesting to study the phenomenological consequences of the coupling term , @xmath17 .
for instance , this term seems to introduce some kind of modified fermion mass , giving rise to an interesting non - trivial effect on the mass renormalization .
s. dimopoulos and h. georgi , nucl .
b193 ( 1981 ) 150 ; see also m. drees , r.m .
godbole and p. roy , _ theory and phenomenology of sparticles : an account of four - dimensional @xmath18 supersymmetry in high energy physics _ , ( world scientific , 2004 ) . | the soft breaking of gauge or other symmetries is the typical quantum field theory phenomenon . in many cases
one can apply the st@xmath0ckelberg procedure , which means introducing some additional field ( or fields ) and restore the gauge symmetry .
the original softly broken theory corresponds to a particular choice of the gauge fixing condition . in this paper
we use this scheme for performing quantum calculations for fermion - torsion theory , softly broken by the torsion mass in arbitrary curved spacetime .
+ g. de berredo - peixoto departamento de fsica , ice , universidade federal de juiz de fora , campus universitrio + juiz de fora , mg 36036 - 330 brazil + [email protected] keywords : softly symmetry breaking , renormalization , propagating torsion , curved space .
+ pacs numbers : 04.62.+v , 11.10.gh , 11.15.-q , 11.30.-j . |
molecular electronics aims at using individual molecules as building blocks of electronic circuits , such as molecule - scale current rectifiers , diodes or memory elements .
thanks to the versatility of synthetic chemistry , custom - made molecules can be made with a wide variety of built - in properties .
magnetic molecules can combine the molecular spin degree of freedom and its interaction with the charge carriers to process and store information , giving birth to the field of molecular spintronics @xcite .
this complementary path to standard spintronics , more focused on using the spin of the charge carriers , proposes to improve the spin coherence length and time thanks to the reduced size of the molecules .
single molecule magnets ( smms ) , with high spin and magnetic anisotropy , are among the most promising magnetic molecules for molecular spintronics .
memory elements and spin filters are some possible applications but also the quantum nature of the spin may allow new functionalities in the field of quantum computation@xcite . from a fundamental research point of view ,
smm in electronic devices are very interesting candidates to study the interplay between charge and spin at the single molecule level . in spite of the extensive knowledge acquired on crystals of smm , there is still limited information of how these individual molecules preserve their magnetic properties when attached to metallic components ( surfaces , electrodes ... ) and how these magnetic properties can be read through their interaction with the electrons .
it has been reported that the fe@xmath0 smm preserves its magnetic properties when grafted onto gold@xcite . moreover ,
these magnetic properties can be quantified and reversibly modified by changing the charge state of the molecule@xcite .
recently , the read - out of the nuclear spin flip@xcite and rabi oscillations@xcite of a tbpc@xmath1 double decker molecule have been reported . here
, we review the main methods used to measure and quantify the magnetic properties of smms , mainly anisotropy and changes in spin , through their interaction with an electrical current in a three - terminal solid state device .
the manuscript is divided as follows . in section [ sec : sect2 ] we describe the magnetic properties of smm with special focuss on the fe@xmath0 smm . in section [ sec : sect3 ] we summarize the different methods to measure transport through magnetic molecules and basic concepts of molecular electronics . in sections [ sec :
sect4 ] and [ sec : sect5 ] we show the detection of the zero - field splitting of a smm in the sequential electron transport and the co - tunneling regime respectively . in sections [ sec :
sect6 ] and [ sec : sect7 ] , we describe the gate spectroscopy technique for axial and transverse magnetic anisotropy . section [ sec : sect8 ] treats the kondo physics in a smm . finally , in sections [ sec
: sect9 ] and [ sec : sect10 ] we discuss excitations that originate from excited spin multiplets and vibrations .
transition - metal based single - molecule magnets are generally made of several magnetic ions coupled together by exchange @xmath2 or superexchange interactions . at low temperatures ( @xmath3 )
the magnetic behaviour can be described by an effective total spin @xmath4 that results from the vectorial addition of the spins ( @xmath5 ) of the individual magnetic ions .
this is known as the giant - spin approximation .
the molecular field defined by the symmetry of the molecule gives rise to a magnetic anisotropy that creates an _ easy magnetization axis _ as sketched in fig.[figure1](a ) .
the degeneracy of the spin ground multiplet breaks into @xmath6 states ( @xmath7 ) as depicted in fig.[figure1](b ) .
the molecule can be modeled by a so - called `` giant - spin hamiltonian '' : @xmath8 where the first term describes the axial anisotropy of the molecule .
the second term characterizes the transverse anisotropy as second order perturbations to the axial anisotropy .
the last term describes the zeeman interaction of @xmath4 with an external magnetic field @xmath9 , where @xmath10 is the land factor and @xmath11 is the bohr magneton .
@xmath12 is the projection of the spin along the easy axis z and @xmath13 is the axial anisotropy parameter . the angle between @xmath9 and the easy axis z is defined as @xmath14 .
the energy levels @xmath15 are distributed over a barrier @xmath16 that separates the `` up '' and `` down '' orientations of the spin ( see fig.[figure1](b ) ) . @xmath17 and @xmath18 are the spin projections on a _ hard _ and _ medium _ axes respectively ( see fig.[figure1](c ) ) and @xmath19 is the transverse anisotropy parameter .
it mixes the @xmath20 states of both sides of the barrier and breaks the degeneracy of the @xmath21 doublets by @xmath22 leading to a small tunnel probability between @xmath20 and @xmath23 states if @xmath4 is integer as sketched in fig.[figure1](d ) . .
the radial distance to the surface represents the energy of a spin as a function of the orientation .
easy _ magnetization axis ( minimum potential ) is defined along the @xmath24 direction .
( b ) spin ground multiplet of an integer spin smm including only axial anisotropy .
( c ) potential energy surface including the transverse anisotropy .
hard _ and a _ medium _ axes for spin reversal are defined .
( d ) spin ground multiplet of an integer spin smm including axial and transverse magnetic anisotropy .
the solid lines connect the states that are mixed by the transverse anisotropy.,scaledwidth=50.0% ] in this review , we focus on the fe@xmath0 smm with formula [ fe@xmath0(l)@xmath1(dpm)@xmath25\cdot$ ] et@xmath1o where hdpm is 2,2,6,6-tetramethyl - heptan-3,5-dione and h3l is the tripodal ligand 2-hydroxymethyl-2-phenylpropane-1,3-diol , which carries a phenyl ring@xcite .
the magnetic core is made of 3 peripheral fe@xmath26 ions ( @xmath27 ) antiferromagnetically coupled to the central fe@xmath26 as schematically shown in fig.[figure2](a ) .
the resulting spin ground state in the neutral charge state @xmath28 is @xmath29 .
the easy axis of the molecule points along the phenyl rings direction , that is , perpendicular to the plane containing the four fe@xmath26 ions as shown in fig.[figure2](b ) .
the first excited spin multiplet ( @xmath30 ) lies around 5 mev above the spin ground multiplet and therefore the giant spin hamiltonian is a good approximation at our experimental temperatures , typically below 1.8 k ( 0.15 mev ) .
the fe@xmath0 smms are not functionalized with any specific group to link to the au electrodes ( like thiol or thiophene ) . the molecule - electrode interaction is therefore mediated only by weak van der waals interactions ideally through the phenyl rings .
three peripheral fe@xmath26 ions are antiferromagnetically coupled to a central fe@xmath26 ion giving a total spin @xmath29 in the neutral charge state @xmath28 .
phenyl rings , peripheral ligands and hydrogen atoms are removed for clarity .
( b ) lateral view of the fe@xmath0 smm showing the phenyl rings that , ideally , contact with the gold electrodes .
_ bottom _ : energy of the spin ground multiplet of the fe@xmath0 smm when @xmath9 is parallel ( c ) and perpendicular ( d ) to the easy axis .
the parameter @xmath14 defines the angle between @xmath9 and the easy axis z.,scaledwidth=50.0% ] magnetic properties of fe@xmath0 smm crystals have been extensively studied with different spectroscopic techniques such as electron paramagnetic resonance epr , neutron scattering and superconducting quantum interference device squid magnetometry . from these measurements
the bulk values of the anisotropy parameters have been determined : @xmath31ev and @xmath32ev yielding a zfs of @xmath33 mev and an energy barrier @xmath34 mev @xcite .
these techniques , however , are only sensitive to large ensembles of molecules and are not valid for the individual smm . as a result of this limitation , little
is known about the magnetic anisotropy of individual and charged smms .
the fe@xmath0 smm , however , has proven to preserve its magnetic properties in solution @xcite when attached to gold surfaces @xcite , electrodes@xcite or nanoparticles@xcite .
we distinguish two different approaches to measure electronic transport through a single molecule depending on the strength of the coupling between the charge carriers and the spins of the smm . in the _ indirect _ approach
, the charge carriers do not flow through the magnetic core of the molecule .
instead , electrons flow through a close - by intermediate transport channel@xcite to which the molecule is linked as shown in fig.[figure4](a ) .
this has been achieved using carbon nanotubes or big organic shells surrounding the magnetic core that are in turn attached to the source / drain electrodes via tunnel barriers .
it is the less invasive of the two approaches and it does not allow to change the redox state of the molecule easily .
in contrast , in the _ direct _ approach described in this paper , the current flows through the magnetic core of the molecule that is directly coupled to the source and drain electrodes via tunnel barriers ( see fig.[figure4](b ) ) .
the scanning tunnel microscope stm configuration depicted in fig.[figure4](c ) is an example of the direct approach .
the tip of the stm is used as the source electrode whereas a metallic substrate acts as drain electrode .
the advantages of the stm approach are the in - situ imaging of the molecule@xcite , the local electrical probing on different parts of the molecule @xcite and a higher control over the electronic coupling @xmath35 . on the other hand ,
a limitation is the absence of a gate electrode .
moreover most molecules hybridize with the surface thereby altering their complex magnetic structure as observed in recent examples@xcite .
this can be solved by using more robust smm like the fe@xmath36 .
smm in the neutral @xmath28 and the reduced @xmath37 charge states .
we only represent the @xmath37 state corresponding to @xmath38 for clarity .
arrows indicate ground to ground ( blue ) , ground to excited ( red ) and excited to excited states ( green ) transitions .
( b ) electrochemical scheme of a first - order sequential tunneling process involving ground states .
( c ) schematic @xmath39 color plot as a function of @xmath40 and @xmath41 .
first order transport is included .
( d ) electrochemical scheme of a first - order sequential tunneling process involving excited states.,scaledwidth=50.0% ] a molecule in the three - terminal devices used hereafter can be schematically depicted as in fig.[figure4](b ) .
the molecule bridges the nanogap formed between source and drain gold electrodes by self - breaking electromigration@xcite and is capacitively coupled to an aluminum gate electrode . by means of the electrostatic field generated by the gate , it is possible to reduce / oxidize the molecule , i.e. , to add / substract one electron to / from the molecule .
the energy needed , for instance , to bring the molecule from its charge state @xmath28 and total spin state @xmath4 to its charge state @xmath37 and total spin state @xmath42 , as shown in fig.[figureset](a ) for the fe@xmath0 , is defined by the electrochemical potential @xmath43 , where @xmath44 and @xmath45 are set by conservation of angular momentum .
the electrochemical potentials of the electrodes can be defined as @xmath46 and @xmath47 for source and drain respectively , where @xmath48 is the fermi level and @xmath40 is a bias voltage .
applying @xmath40 and gate @xmath41 voltages , the relative alignment between the electrochemical potential of the electrodes and the molecular states @xmath49 , @xmath50 can be varied .
an electrochemical scheme of the molecule and the electrodes is shown in fig.[figureset](b ) .
if the condition for resonant transport is met ( @xmath51 ) electrons can hop one at a time from the source to the molecule and from the molecule to the drain , in a two - step process .
transport in this regime is denominated single - electron tunneling ( set ) transport and takes place according to the cycle @xmath52 . upon reaching the resonant transport condition , the differential conductance ( @xmath39 ) shows a peak and the current through the molecule has a step - like increase .
a typical representation of the result of a measurement is shown in fig.[figureset](c ) ; it is a color plot of the @xmath39 as a function of @xmath40 and @xmath53 .
the red slanted lines , known as coulomb edges , are due to transitions from the ground state of one charge state to the one of an adjacent charge state .
the crossing point at zero bias is often called degeneracy point or coulomb peak . in the case of the fe@xmath0 smm
, the transition may correspond to the ( @xmath54 , @xmath55 ) and ( @xmath56 , @xmath57 ) or ( @xmath58 , @xmath59 ) marked with a blue arrow in fig.[figureset](a ) .
first - order tunneling transport is energetically allowed in the high - bias regions of the conductance map confined by those edges , indicated by set .
we have measured around 300 junctions at low temperatures ( @xmath60 k ) .
most of them do not show any gate dependence or are very high ohmic junctions ( @xmath61 ) indicating that the junctions are empty .
interestingly , 31 junctions exhibited gate dependence or steps in @xmath62 measurements . from these , 16 displayed a clear coulomb peak separating two stable charge states@xcite .
the state - of - the - art gating technology with solid gates is limited by the maximum voltage supported by the gate and the dielectric and the efficiency of the coupling @xmath63 between the electrical field and the molecule .
the value of @xmath63 is experimentally obtained as @xmath64 where @xmath65 and @xmath66 are respectively the positive and negative slopes of the coulomb edges .
typical values are @xmath67 which allow to shift the energy levels by hundreds of mev . given the fact that charging energies of small molecules are expected to be of the order of ev , @xmath39 color maps therefore are expected to show two stable charge states separated by an set region .
the position of the degeneracy point is not a priori determined as it shifts in gate voltage from sample to sample what makes difficult to assign the neutral charge state to a specific coulomb diamond .
the different magnetic behavior of the different charge states can be still used , in some cases , to assign the neutral charge state as explained in the next sections .
the electronic coupling of the molecule to the source @xmath68 and drain @xmath69 electrodes defines the rate at which the electrons tunnel through the molecule .
@xmath35 can be then defined as @xmath70 .
at very low temperatures ( @xmath71 ) this electron coupling is the main contribution to the broadening of the coulomb edges and peak .
the value of @xmath35 can therefore be estimated from the fwhm of the coulomb peak at zero bias according to @xmath72 .
this value depends on the specific arrangement of the molecule between the electrodes and can not be controlled in solid - state devices where the electrodes are fixed . in the case of the fe@xmath36 smm
, we found that @xmath35 ranges from 0.5 mev to 3.8 mev in the different samples .
it is worth noting that we observe asymmetric coupling ( @xmath73 ) in @xmath74 of the measured fe@xmath36 smm junctions .
excited states can contribute to electron transfer once their chemical potential ( @xmath75 ) lies within the bias window , see red arrows in fig.[figureset](d ) . depending on the initial and final states ,
they are referred to as ground - to - excited , excited - to - ground ( red arrows in fig.[figureset](a ) ) or excited - to - excited ( green arrow in fig.[figureset](a ) ) state transitions and follow cycles of the kind @xmath76 , where @xmath77 and @xmath78 denote initial and final @xmath79 states that fulfill @xmath80 and @xmath81 is a state that fulfills @xmath82 .
these transitions manifest themselves as additional lines running parallel to the transport edges within the set regions that intersect the opposite coulomb edges at finite bias ( see fig.[figureset](c ) ) .
the intersecting bias gives the energy of the excitation @xmath83 with respect to the ground state .
moreover , from the direction and position of these transport lines , it is possible to determine to which charge state the excitations belong .
for instance , the transport lines terminating in the @xmath28 charge - state coulomb diamond indicated in fig.[figureset](c ) are related to the energy difference @xmath84 between two @xmath15-spin states belonging to the spin @xmath4-multiplet of the @xmath28 charge state .
colour plot of an individual fe@xmath36 smm junction as a function of @xmath40 and @xmath85 at @xmath86 t and @xmath87 mk ( sample ab ) .
coulomb - blockade is visible on either side of the degeneracy peak , marking two regions with a definite charge state .
faint inelastic co - tunneling that can be attributed to the zfs lines are also visible . within the set transport regions , excitations of vibronic origin appear .
the asymmetry in the couplings to the electrodes is evident in the different intensities of the coulomb edges .
( b ) same colour plot as in ( a ) measured at @xmath88 t and @xmath87 mk .
the two coulomb edges at positive and negative biases ( black arrows ) together with the corresponding zfs inelastic cotunneling excitations are now split .
( c ) @xmath89 colour plot calculated with a three - state master equation model at @xmath86 t. ( d ) same as in ( c ) calculated at @xmath88 t. in both simulations , the different conductance intensities of the coulomb edges can be reproduced with a ratio @xmath90 , where the prime indicates excited state related quantities . ] in fig.[figure5](a ) the @xmath89 map of an fe@xmath0 junction ( sample ab ) as a function of @xmath40 and @xmath91 measured at @xmath87 mk and a magnetic field @xmath92 t is shown .
two low - conductance regions on either side of the degeneracy point at gate voltage @xmath93 v can be distinguished .
these two coulomb - blockaded regions can be associated with the two stable states ( @xmath94 ) and ( @xmath95 ) for the lower and higher gate voltages respectively ( determination of the states will be explained in sect.[sec : sect6 ] ) .
figure [ figure5](b ) shows the conductance map at the same temperature and at an applied magnetic field @xmath96 t. the edges marking the transition between the set transport and the @xmath37 charge state have split in two pairs of lines ( indicated by arrows ) .
the excited - to - ground state excitation terminates now at @xmath97 mev .
the increase of the excitation energy is consistent with the evolution of the zero - field splitting in a magnetic field . assuming @xmath98 mev as in bulk , the splitting of the lines and the energy @xmath99 mev can be reproduced for @xmath100 in equation ( [ gs ] ) .
this angle is confirmed by gate spectroscopy measurements performed on the sample ab that are described in sect.[sec : sect6 ] .
note that because of the lifetime broadening @xmath101 the zfs can not be resolved at low fields .
the analysis of the energies and the allowed spin transitions thus let us conclude that this transition follows the cycle @xmath102 . + in order to confirm that the observed excitation corresponds to the zero - field splitting transition , differential conductance maps at various magnetic fields are simulated with a minimal three - state rate equation model .
the magnetic energy levels are derived by diagonalizing the hamiltonian in equation ( [ gs ] ) for the two charged states @xmath28 and @xmath37 and for various magnetic fields .
transition rates , occupation probabilities of the magnetic levels and the resulting current are then calculated within the standard ( stationary ) master equation framework , assuming thus weak coupling between the molecule and electrodes as expected for a physisorbed molecule .
motivated by the low thermal energy and the weak mixing of the spin eigenstates at angles @xmath103 , the model assumes the magnetic number @xmath20 to be an approximately good quantum number and includes only the first two lowest spin eigenstates of each charge state .
the results of the calculations for @xmath86 t and @xmath88 t at an angle @xmath104 are shown in fig.[figure5](c , d ) respectively .
as can be seen , the energy of the transition @xmath105 increases in field consistently with the experimental data . in order to reproduce the specific conductance values observed in the experiments , asymmetric tunnel couplings to the electrodes for the states
are employed .
good agreement is obtained for a ratio @xmath106 .
an asymmetrical geometric arrangement of the molecule with respect to the source and drain electrodes can account for this ratio .
+ additional transport lines present in the set are absent in the simulations .
these excitations can be attributed to vibron - assisted processes , as they are roughly harmonically - spaced and do not change their energy in magnetic field .
further details on the interaction with vibrational states is given in sect.[sec : sect10 ] .
magnetic excitations can also contribute to transport in the coulomb blockade regime@xcite . in this case ,
sequential electron transport is suppressed and therefore the magnetic transitions can only take place via high - order processes like inelastic co - tunneling .
the current level for a co - tunneling process is proportional to @xmath107 and therefore it is only relevant for intermediate or high values of @xmath35 ; i.e. , @xmath108 ( see ref.@xcite for a review ) . .
( b ) stability diagram of a magnetic molecule showing different high - order tunneling processes .
( c ) spin ground multiplet for the fe@xmath36 smm showing the magnetic transition corresponding to an inelastic co - tunneling process .
( d ) electrochemical scheme of a kondo spin - flipping process.,scaledwidth=50.0% ] figure [ figure3](a ) shows the electrochemical scheme of an inelastic co - tunneling process .
one electron hops from the molecule into the drain leaving the molecule in a forbidden virtual state .
this event is possible if another electron jumps into the molecule in a time @xmath109 according to heisenberg s uncertainty principle .
there , @xmath110 is the charging energy .
the incoming electron can jump into a different excited state with energy @xmath83 as long as @xmath111 .
a step in @xmath39 appears at the threshold voltage @xmath112 as sketched in fig.[figure3](b ) that merges with the corresponding set excitation at the diamond edge .
two electrons are involved in this second - order process and the spin selection rules dictate that only transitions fulfilling @xmath113 and @xmath114 are allowed . in the fe@xmath0 smm , the transition from the ground state ( @xmath54 , @xmath55 ) to the first excited state ( @xmath54 , @xmath115 ) is the most prominent transition in the co - tunnel regime , as sketched in fig.[figure3](c ) ; i.e. , it concerns the zfs in a single charge state where @xmath116 and @xmath117 .
the zfs splitting is the largest energy difference in the spin ground multiplet @xmath54 .
accordingly , once the transition is activated , all other transitions are allowed as well and therefore only one step in the @xmath39 is expected . at higher bias voltages , transitions to excited spin multiplets fulfilling @xmath118 ( @xmath54 to @xmath30 in the neutral state ) may also become visible as discussed in sect.[sec : sect9 ] .
figure [ figurecot](a ) shows the @xmath89 color plot measured at @xmath119 mk as a function of @xmath40 and @xmath85 of an fe@xmath0 junction ( sample ah ) showing again two adjacent charge states .
two parallel co - tunneling lines show up within the left - hand charge state with energy @xmath120 mev that is close to the expected energy of the zfs in bulk fe@xmath0 .
two symmetric excitations appear as well within the right charge state with energy @xmath120 mev . to gain a deeper insight in the magnetic nature of the excitations ,
the evolution of @xmath83 in a magnetic field has been examined .
color plot as a function of @xmath40 and @xmath91 for an fe@xmath0 smm junction ( sample ah ) measured at @xmath119 mk .
co - tunneling excitations appear at 0.5 mv in the left and right - hand charge states .
( b ) @xmath89 traces measured at @xmath121 v and different vales of @xmath9 . the energy of the excitation increases with increasing @xmath9 as expected for the zfs .
solid lines are fits to a lambe - jaklevic formula to obtain the energy of the excitation @xmath83 .
( c ) and ( d ) @xmath83 plotted as a function of @xmath9 for the left and the right charge states respectively .
the solid lines are fits to equation ( [ gs ] ) with parameters @xmath122ev , @xmath29 , @xmath123 and @xmath124ev .
orange and blue dashed lines are upper and lower boundaries for @xmath125 and @xmath126 . ,
scaledwidth=50.0% ] figure [ figurecot](b ) shows different @xmath89 traces measured at @xmath127 v and at different magnetic fields .
the co - tunneling lines clearly move apart by increasing the magnetic field ; i.e. , the energy of the excitation increases as expected for the @xmath83
. the solid lines are fits of these curves to a lambe - jaklevic equation@xcite : @xmath128 where the parameters @xmath129 and @xmath130 , are horizontal and vertical offsets respectively , and @xmath131 and @xmath132 account for bias asymmetries in the @xmath89 .
the energy @xmath83 obtained from these fits is plotted as a function of @xmath9 in figs.[figurecot](c ) and ( d ) for the left and the right - hand charge state respectively .
the evolution of @xmath83 in a magnetic field provides information about the magnetic anisotropy of the molecule@xcite since the zfs is very sensitive to @xmath14 and @xmath13 as seen in fig.[figure2](c - d ) .
the solid line in fig.[figurecot](c ) is a fit to @xmath83 calculated by numerical diagonalization of equation ( [ gs ] ) with @xmath29 @xmath122ev and @xmath123 , @xmath133 . the same analysis is performed for the right charge state in fig.[figurecot](d ) and yields @xmath124ev and @xmath134 .
the magnetic anisotropy therefore increases around @xmath135 by adding one electron to the molecule .
the nonlinear dependence of the @xmath83 with @xmath9 is proof of the magnetic anisotropy of the molecule and points to high values of @xmath14 .
this nonlinear dependence has also been observed in samples ab and b. the orange and blue dashed lines in figs.[figurecot](c ) and [ figurecot](d ) are upper and lower boundaries of the fit by using different values of @xmath136 and @xmath125 respectively .
note that the high-@xmath9 part of the dependence is more sensitive to variations in @xmath14 whereas the low-@xmath9 part is more sensitive to variations in @xmath13 .
this is a consequence of the predominance of the zeeman effect over the axial anisotropy at high @xmath9 and viceversa .
we have observed low - bias co - tunneling excitations in 7 of the 31 junctions showing molecular signatures .
the values of the excitations are summarized in table.[tablezfscot ] .
we generally find that the values are of the order of 0.5 mev with variations from sample to sample .
typical values for the zfs are lower than 1 mev ( 0.55 mev in the case of bulk fe@xmath36 ) .
additional low - bias excitations , such as zero - bias anomalies or vibrations , are often present in the measurements depending on the coupling of the molecule with the electrodes .
the identification of the zfs may therefore not be straightforward in those cases .
figure [ kondo](c ) illustrates one of these examples . in that case
, the zfs appears as small shoulders on top of a zero - bias resonance due to kondo interactions as explained in sect.[sec : sect8 ] .
.statistics of low - bias co - tunneling excitations ( @xmath137 mev ) .
the values of @xmath83 of samples with + are obtained from the fits to the lambe - jaklevic equation at @xmath92 t. for the other two , @xmath83 could not be obtained at low @xmath9 , e.g. masked by a kondo peak ( sample y in fig.[kondo](c ) ) or not visible at @xmath92 ( sample ab in fig.[figure5](a ) ) .
@xmath138 is then obtained by extrapolating high-@xmath9 values of @xmath83 to @xmath92 t. samples a and b were previously reported in ref.@xcite .
sample g was reported in ref.@xcite . [ cols="^,^,^",options="header " , ] the behavior of the kondo anomaly so far is in good agreement with a @xmath139 kondo model .
interestingly , several theoretical works predict a more exotic kondo arising from the magnetic anisotropy and high spin of smm ; i.e. , high - spin under - screened kondo and quantum - spin tunneling kondo @xcite .
under - screened kondo has been observed in a c@xmath140 molecule@xcite , but its experimental observation in smm with larger spin still remains an open challenge to the best of our knowledge .
high - energy excitations , with energies that appear beyond the ground spin multiplet of the fe@xmath36 smm ( @xmath141 mev ) , show up in transport ( set and co - tunneling ) when triggered by higher bias voltages . in principle
, excitations can be either of magnetic or vibrational origin .
magnetic excitations can be identified due to their characteristic magnetic field evolution in contrast with vibrational excitations for which the energy is expected to be magnetic field independent . figure [ figure7](a ) shows the stability plot measured at @xmath142 mk for an fe@xmath36 junction showing a strong excitation at 4.8 mev in the set that merges in the right diamond edge with the correspondent co - tunneling excitation coming from the right - hand charge state .
the energy of the excitation approximately matches the energy in bulk for the @xmath30 excited multiplet of the fe@xmath36 @xcite and appears , depending on @xmath35 , as set or co - tunneling excitations in 8 of the measured samples .
figure [ figure7](b ) shows a @xmath39 trace taken at @xmath143 v corresponding to the coulomb blockade regime in fig.[figure7](a ) .
a step in the conductance shows up which is symmetric by reversing the bias polarity .
the solid line is a fit to the lambe - jaklevic equation ( [ lj ] ) with 4.8 mev .
an additional example showing co - tunneling excitations at around 5 mev is shown in figs.[figure7](c ) and ( d ) for a different junction containing fe@xmath36 . in order to confirm the magnetic origin of these high - bias excitations , further magnetic field measurements are necessary and will by the subject of a further study .
color map measured versus @xmath40 and @xmath85 in an fe@xmath36 junction ( sample ac ) measured at @xmath142 mk . a clear set excitation shows up at 5 mev that merges at the coulomb edge with the correspondent co - tunneling excitation in the right - hand charge state .
( b ) @xmath89 trace measured at @xmath144 v in the coulomb blockade regime of ( a ) .
an inelastic co - tunneling excitation shows up as step in the @xmath89 centered at @xmath145 mev .
the solid line is a fit to the lambe - jaklevic equation .
( c ) @xmath89 color map of a different fe@xmath36 junction ( sample p ) and ( d ) @xmath89 trace at v@xmath146 v. an inelastic co - tunneling process shows up centered around @xmath145 mev.,scaledwidth=50.0% ] according to the spin selection rules described in sect.[sec : sect4 ] and [ sec : sect5 ] , the set excitation would correspond to a @xmath147 transition in the set regime and a @xmath148 in the coulomb blockade .
interestingly , the same selection rules forbid the set transition involving the excited multiplet of the @xmath149 charge state , that is @xmath150 , since @xmath151 .
only one magnetic excitation should be visible in the @xmath89 color plot ending in the coulomb edge that delimits the @xmath54 neutral state whereas it should be absent in the coulomb edge that delimits the @xmath38 charged state .
this asymmetry combined with the presence of zfs and the sign of @xmath152 slope in gate spectroscopy , may be used to determine the neutral charge state ; i.e. , the right - hand one in the specific case shown in figs.[figure7](a , c ) or the left in fig.[kondo](a ) .
this last example ( sample f ) is consistent with the positive slope of @xmath153 found by gate spectroscopy which also indicates that the left charge state is the neutral one@xcite .
vibrations are present in all kind of nanostructures and can play an important role in their mechanics . in addition , vibrational modes can couple to the charge carriers and therefore show up in conductance measurements .
figure [ figurefc](a ) shows the @xmath89 color plot of an fe@xmath0 junction in which vibrations appear as set lines parallel to the diamond edges@xcite .
the measurements are performed at @xmath154 k. the excitations can also be observed in the conductance plot in figs.[figure5](a ) and [ figure5](b ) .
the energies of vibrational excitations are independent of the magnetic field in contrast with magnetic excitations . moreover , the energy is also independent of the bias polarity and the charge state .
figure [ figurefc](b ) shows the numerical derivative of the @xmath89 color plot in fig.[figurefc](a ) in which vibrational excitations become more visible .
color plot as a function of @xmath40 and @xmath91 for an fe@xmath0 junction ( sample l ) measured at @xmath154 k. several equally - spaced excitations and a low - bias suppression of the set are observed .
these are fingerprints of franck - condon blockade due to a high electron - phonon coupling .
this figure is adapted from ref.@xcite .
( b ) numerical derivative of @xmath89 color plot shown in ( a ) .
vibrational excitations become more visible .
( c ) franck - condon factors for different electron - phonon couplings @xmath155 .
higher harmonics @xmath156 are excited for large @xmath155 values .
( d ) @xmath39 trace at a fixed @xmath91 .
this figure is adapted from ref.@xcite , scaledwidth=50.0% ] .
[ figurefc ] the strength of the electron - phonon coupling @xmath155 determines the @xmath39 height of the excitations ( @xmath157 ) . in short
, @xmath157 is proportional to the franck - condon factors @xmath158 where @xmath156 is the quantum state of the vibration in an harmonic oscillator approximation .
figure [ figurefc](c ) shows the franck - condon factors for different values of @xmath155 for transitions between the ground vibrational state and higher harmonics .
intermediate and high values of @xmath155 excite higher harmonics of the vibration that appear in the conductance when the bias matches a quantum of the energy of the vibration @xmath159 .
this is the case in fig.[figurefc](b ) where at least @xmath160 harmonics can be observed .
the regular energy spacing of the excitations indicates that a single vibrational mode is the origin of the spectrum .
the energy spacing , taken from the intersection of the excitations with the diamond edge , is the energy of the vibrational mode and equals @xmath161 mev @xcite . even if other vibrational modes of the molecules are excited , only those with the higher @xmath155 will contribute significantly to the current @xcite . for higher values of @xmath155 , the fc factor for ground state to ground state transitions ( @xmath162 ) decreases exponentially and becomes negligible as seen in fig.[figurefc](c ) .
low - bias set transitions are therefore suppressed and a low - bias gap appears in the conductance that can not be lifted with a gate voltage , as shown in fig.[figurefc](a ) . only by increasing @xmath40 ,
higher harmonics become available enabling transport and lifting the blockade . this vibrational - induced suppression of the current
is known as franck - condon blockade .
the value of @xmath155 can be estimated from the relative height of the different harmonics of the oscillation .
figure [ figurefc](d ) shows a @xmath39 trace measured at a fixed @xmath163 v. the background contribution due to direct tunneling in the junction is subtracted from the data . the solid line is a fit to the franck - condon factor @xmath164 with @xmath165 .
an independent estimate of @xmath155 can be obtained from the size of the gap as @xmath166 where @xmath167 is the threshold bias voltage that restores set .
the value obtained from fig.[figurefc](a ) is @xmath168 .
the value of @xmath155 may vary from sample to sample due to environmental factors like image charges and different molecular orientations relative to the electrodes @xcite .
this may explain why the franck - condon blockade is not present in all junctions .
we have observed vibrational excitations in 10 samples and a low - bias suppression of the current in three of them .
vibrational energies are in the order of 2 mev .
we further note that our measurements show that @xmath155 can be high in smm .
it is unclear how such a strong electron - phonon coupling affects the magnetic properties of the fe@xmath0 .
for instance , vibrations are known to be a source of spin decoherence@xcite and may lead to quantum interferences@xcite .
we have reviewed different methods to detect and measure magnetic properties of single molecules in electronic transport .
we have focussed on the fe@xmath36 smm and provided statistics of the relevant magnetic properties of the molecule .
we find that the coupling of the molecules to the electrodes @xmath35 , the gate coupling parameter @xmath63 and the electron - phonon coupling vary from sample to sample which may be explained by a change in the arrangement of the molecule between the electrodes .
this variation induces changes in the perceptibility of the magnetic and vibrational excitations or kondo anomalies .
however , the magnetic properties of the molecule like the high - spin and the axial magnetic anisotropy are preserved , including a reversible increase in the magnetic anisotropy upon charging of the molecule .
these results point to the fe@xmath36 smm as a good platform to study the interesting physics arising from the spin - charge interaction at the single - molecule level .
quantum effects , such as quantum tunnel or quantum oscillations induced by the transverse anisotropy may have an impact on the current through the molecule .
in addition , further studies are needed to shed more light into the spin - vibration interaction or the existence of exotic kondo behavior due to the high spin of the molecule .
accorsi s , barra a - l , caneschi a , chastanet g , cornia a , fabretti a c , gatteschi d , mortal c , olivieri e , parenti f , rosa p , sessoli r , sorace l , wernsdorfer w , zobbi l 2006 _ j. am .
* 128 * 4742 - 55 . | we review different electron transport methods to probe the magnetic properties , such as the magnetic anisotropy , of an individual fe@xmath0 smm .
the different approaches comprise first and higher order transport through the molecule .
gate spectroscopy , focusing on the charge degeneracy - point , is presented as a robust technique to quantify the longitudinal magnetic anisotropy of the smm in different redox states .
we provide statistics showing the robustness and reproducibility of the different methods .
in addition , conductance measurements typically show high - energy excited states well beyond the ground spin multiplet of smm . some of these excitations have their origin in excited spin multiplets , others in vibrational modes of the molecule .
the interplay between vibrations , charge and spin may yield a new approach for spin control . |
in the study of the static properties of finite systems the main quantity of interest is the density of states @xmath0 as it contains the complete physical information of the investigated system @xcite . to be specific , consider a magnetic system where @xmath1 is a function of the energy @xmath2 and of the magnetization @xmath3 ( which can be a scalar or a vector quantity , depending on the system ) . with the density of states at hand
one may proceed in the canonical way by computing the free energy from which other quantities such as the susceptibility of the specific heat can be derived . in the last years , however , an increasing number of groups have chosen to directly investigate the density of states or , equivalently , the microcanonical entropy @xmath4 ( setting @xmath5 ) . the origin of this increasing interest in the microcanonical approach can be found in the observation that ensembles are not always equivalent @xcite .
the inequivalenve of ensemble is for example encountered in situations where the macroscopic limit ( for example due to the presence of long - range interactions such as the gravitational force ) does not exist . but also in finite systems with short - range interactions the canonical and the microcanonical ensembles may give different results although the ensembles have to become equivalent in the infinite volume limit .
this is especially the case for finite systems which in the macroscopic limit undergo a phase transition .
it is the purpose of this paper to discuss the finite - size signatures of phase transitions in the microcanonical ensemble @xcite .
the paper is organized in the following way . in section 2
we recall that in the microcanonical approach all quantities of interest can exclusively be expressed as derivatives of the entropy @xcite . in section 3
we briefly discuss the case of discontinuous phase transitions which lead to a characteristic behaviour of the specific heat in finite microcanonical systems .
section 4 is the main part of this paper .
it is devoted to the signatures of continuous phase transitions encountered in the microcanonical analysis of small systems .
we also discuss different ways for extracting the values of critical exponents directly from the derivatives of the microcanonical entropy .
conclusions are given in section 5 .
we use in the following the language of magnetism and consider a magnetic system with @xmath6 degrees of freedom isolated from any environment . the state of the system is then characterized by its energy @xmath2 and its magnetization @xmath3 . in the microcanonical analysis one
considers the microcanonical entropy density @xmath7 as function of the energy density @xmath8 and the magnetization density @xmath9 .
the entropy surface @xmath10 still exhibits a weak size dependence which is not explicitly denoted here .
sometimes one is only interested in the energy dependence and therefore studies the reduced specific entropy @xmath11 with @xmath12 microcanonical quantities can be expressed by functions of derivatives of the specific microcanonical entropy .
for example , the microcanonical temperature reads @xmath13 which leads to the following expression for the specific heat : @xmath14 it is worth noting that the microcanonical temperature and specific heat can also be defined through the entropy density ( [ eq : s ] ) .
these differently defined quantities get identical in the infinite volume limit .
we refer the reader to @xcite for a thorough discussion of this point .
further quantities discussed in the following include the spontaneous magnetization and the susceptibility .
the spontaneous magnetization @xmath15 is defined to be the value of @xmath16 where the entropy ( [ eq : s ] ) at a fixed value of @xmath17 has its maximum with respect to @xmath16 @xcite : @xmath18 to better understand this definition consider the density of states @xmath19 for a given energy @xmath17 .
the density of states exhibits a sharp maximum as the overwhelming majority of accessible states belongs to the value @xmath20 of @xmath16 where @xmath21 has its maximum .
this is the reason why we identify @xmath20 with the spontaneous magnetization of the finite system .
this definition of the microcanonical spontaneous magnetization ensures that we recover the canonically defined spontaneous magnetization in the macroscopic limit .
note also that one can define a magnetic field for the microcanonical ensemble which appears as the conjugate variable to @xmath16 . at the spontaneous magnetization
@xmath20 the associated magnetic field becomes zero , see @xcite for further details .
of further interest is the suceptibility @xcite @xmath22 where the denominator on the right hand side is in fact the local curvature of the entropy surface .
as it stands , eq .
( [ eq : chi ] ) is only valid in cases where the order parameter is a scalar . for an order parameter with vector character ( as encountered for example in classical spin models like the @xmath23 or the heisenberg model ) the susceptibility is given by a matrix @xcite .
the entries of the susceptibility matrix are then dependent on the local curvature with respect to the different order parameter components . closing this section
, we reemphasize that the microcanically defined quantities all fulfill the important request that they become identical to the better known canonical quantities in the infinite system after the trivial transformation from the natural variables @xmath17 and @xmath16 of the microcanonical description to the natural variables temperature @xmath24 and magnetic field @xmath25 of the canonical approach .
the interest in the microcanonical analysis really started with the observation that discontinuous phase transitions display typical and very interesting finite - size signatures in the microcanonical treatment @xcite .
indeed , a discontinuous phase transition reveals itself in finite systems by a convex intruder in the microcanonical entropy which originates from states of coexistence of different phases . in the canonical treatment these states are unstable and can not be accessed by equilibrium methods .
interestingly , this convex intruder leads to a characteristic backbending of the caloric curve and thus to a negative specific heat .
the appearance of a negative specific heat may look like an oddity of the microcanonical approach .
however , there have been recent claims that negative specific heats had been measured ( even so in a somewhat indirect way ) in experiments on nuclear fragmentation @xcite or on the solid - liquid transition of atomic clusters @xcite .
the main intention of this paper is to discuss the typical finite - size signatures of continuous phase transitions encountered in the microcanonical approach .
we thereby focus on classical spin systems and shall discuss in the following the two- and three - dimensional ising models on hypercubic lattices with periodic boundary conditions .
the ising model is defined by the hamiltonian @xmath26 where the lattice site @xmath27 is characterized by the classical variable @xmath28 .
the sum in equation ( [ eq : ising ] ) is over nearest neighbour pairs .
the two - dimensional three - state potts @xcite and the three - dimensional @xmath23 @xcite models have also been studied recently in the present context , but we will not discuss these models here .
all the mentioned models have in common that the corresponding infinite system displays a continuous phase transition separating a paramagnetic phase at high temperatures ( energies ) from a ferromagnetically ordered phase at low temperatures ( energies ) . in the canonical ensemble
the typical features of spontaneous symmetry breaking like the abrupt onset of the order parameter or a diverging susceptibility are only encountered in the infinite system . in any finite system one
does not observe diverging quantities but only rounded maxima .
this is , however , completely different in the microcanonical ensemble where the features of spontaneous symmetry breaking turn up already for finite systems @xcite .
it is instructive to first inspect directly the entropy surface @xmath10 .
figure [ fig_1 ] displays this entropy surface for a three - dimensional ising model with @xmath29 spins .
the fully magnetized states at @xmath30 and @xmath31 form the two degenerate gound states , whereas the fully disordered macrostate at @xmath32 has the highest degeneracy of all .
this entropy surface has been computed by a very efficient numerical method based on the concept of transition variables @xcite .
more insights are gained by looking at cuts through the entropy surface along the magnetization at a fixed energy @xmath17 , see figure [ fig_2 ] .
one observes that the entropy exhibits only one maximum at @xmath33 for energies @xmath17 above an energy @xmath34 , whereas for energies @xmath35 two maxima are seen at nonzero magnetizations , separated by a minimum at @xmath33 . at the energy @xmath36 the curvature in magnetization direction vanishes at @xmath33 .
= 6.0 in = 4.0 in recalling that the spontaneous magnetization is identified with the magnetization where for a fixed energy the density of states is the largest , one can directly extract the value of the spontaneous magnetization at a given energy from cuts like those shown in figure [ fig_2 ] .
the resulting behaviour of the spontaneous magnetization is discussed in figure [ fig_3 ] for three - dimensional ising models containing @xmath29 and @xmath37 spins .
two different regimes are observed : for @xmath38 the spontaneous magnetization is zero , whereas for @xmath35 it is given by @xmath39 , in accordance with the appearance of two maxima of the entropy surface at lower energies . at the well - defined size - dependent energy @xmath36 the order parameter sets in abruptly already in finite systems .
the energy @xmath36 may therefore be called transition energy .
= 4.0 in another remarkable finite - size signature of continuous phase transitions is shown in figure [ fig_4 ] .
a divergence of the magnetic susceptibility is observed for finite systems at the same energy @xmath36 at which the order parameter bifurcates .
this divergence is in fact readily understood by recalling , see equation ( [ eq : chi ] ) , that the susceptibility is proportional to the inverse of the curvature of the entropy surface .
it is the existence of a point of vanishing curvature that is responsible for the observed divergence of the susceptibility , see figure [ fig_2 ] .
= 4.0 in the microcanonical specific heat , on the other hand , does not diverge in finite systems .
it nevertheless also displays an intriguing finite - size behaviour , as revealed recently for models with a diverging specific heat in the infinite system @xcite .
indeed , in the microcanonical ensemble the maximum of the specific heat of the corresponding finite systems varies non - monotonically when increasing the size of the system .
starting with small systems , it first _ decreases _ for increasing system sizes , before it increases for larger systems and finally diverges in the macroscopic limit .
this behavious is in strong contrast to the treatment in the canonical ensemble where a monotonic increase of the maximum of the specific heat with the size of the system is always observed @xcite . in @xcite
a phenomenological theory has been proposed that acccounts for the observed non - monotonic variation of the maximum of the microcanonical specific heat . at this stage one
may wonder whether the typical features of spontaneous symmetry breaking observed in finite microcanonical systems are described by the true critical exponents of the infinite system .
a rapid inspection of the numerical data reveals that this is not the case .
indeed , the onset of the order parameter or the divergence of the suceptibility in finite microcanonical systems are in all cases governed by the classical mean - field exponents @xcite .
this is readily understood by noting that the entropy surface can be expanded in a taylor series in the vicinity of @xmath36 ( for a discrete spin model a suitable continuous function that represents the data has to be chosen ) . as a consequence one ends up with a landau - like theory @xcite which for the microcanonical entropy of a finite system is in principle exact , thus yielding classical values for the critical exponents . in recent years
we have developed different approaches that enable us nevertheless to obtain the true critical exponents by exclusively analyzing the density of states of finite systems @xcite . before discussing these approaches for the spontaneous magnetization in some detail we have first to pause briefly in order to recall that in the entropy formalism considered here critical exponents often differ from their thermal counterparts .
this is the case when the specific heat diverges algebraically at the critical point of the infinite system , implying that the critical exponent @xmath40 , governing the power - law behaviour of the canonical specific heat in the vicinity of the critical point , is positive .
it is then easily shown that the critical exponents @xmath41 appearing in the microcanonical analysis are related to the thermal critical exponents @xmath42 by @xmath43 @xcite .
however , when the specific heat does not diverge but presents a cusp - like singularity as @xmath44 , we have @xmath45 @xcite . in the case of a logarithmically diverging specific heat
the thermal and microcanonical exponents are also identical .
let us now come back to the determination of the critical exponents . as already mentioned we obtain classical values for the critical exponents when expanding the finite - size quantities around the energy @xmath36 .
the spontaneous magnetization , for example , displays a root singularity for energies @xmath35 close to @xmath36 : @xmath46 where the amplitude @xmath47 depends on the system size . in the infinite system one
expands around the true critical point @xmath48 , leading to the asymptotic behaviour @xmath49 for small @xmath50 with @xmath51 , where @xmath52 denotes the ground state energy per degree of freedom . the critical exponent @xmath53 is given by @xmath54 with @xmath55 being the usual thermal critical exponent characterizing the singularity of the spontaneous magnetization .
here we suppose that @xmath56 , as it is the case for the ising model .
away from the critical point the spontaneous magnetization of the infinite system is not any more given by the simple expression ( [ eq : mexpinf ] ) , as correction terms get important .
one way to deal with these correction terms is to analyse the energy dependent effective exponents @xmath57 that yields the true critical exponent @xmath53 in the limit @xmath58 @xcite . in order to use equation ( [ eq : beff ] )
we need data free of finite - size effects . on the other hand one
observes in figure [ fig_3 ] that there exists a energy interval for which the spontaneous magnetizations computed in systems of different sizes are identical .
this suggest the following procedure for computing @xmath53 .
calculate the spontaneous magnetization for a series of increasing system sizes @xmath59 . in the energy range where @xmath60 and @xmath61 yield the same exponent @xmath62 , select this exponent for the infinite system , until finite - size effects show up . then plot the common exponent for system sizes @xmath61 and @xmath63 until these begin to disagree and so on .
this procedure is rather cumbersome but ensures that data essentially free of finite - size effects are obtained .
the result of this approach @xcite for the two- and the three - dimensional ising models is shown in figure [ fig_4 ] .
for the three - dimensional case we observe that @xmath62 rapidly approaches a constant value that perfectly agrees with the expected asymptotic value @xmath64 . in two dimensions
the agreement is not as perfect , even so a reasonable estimate of @xmath53 is obtained by linearly extrapolating the existing data .
the situation is indeed more complicated in the two - dimensional ising model , as one is dealing here with the rather pathological case of a logarithmically diverging specific heat with @xmath65 .
in fact , we expect for any model with @xmath66 a similar good agreement between the expected value of @xmath53 and the value obtained from linearly extrapolating @xmath67 as for the three - dimenisonal ising model .
this expectation has recently been confirmed for the three - dimensional @xmath23 model @xcite .
= 4.0 in the determination of critical exponents via effective exponents has the major drawback that data free of finite - size effects have to be obtained , which means that systems of increasing sizes have to be simulated on approach to the critical point . for figure [ fig_5 ]
our largest systems contained @xmath68 spins in two and @xmath69 spins in three dimensions .
simulating these systems is very time consuming even with the best algorithms available .
the second approach for determining the values of critical exponents discussed in the following does not need large systems , but instead it explicitly takes advantage of the finite - size effects . in the canonical ensemble finite - size scaling theory
is a valuable and often used tool for extracting critical exponents from finite - size data .
its starting point is the observation that in the asymptotic limit @xmath70 and @xmath71 ( @xmath72 being the critical temperature ) the behaviour of finite - size quantities is governed by scaling functions that are basically determined by the ratio @xmath73 with @xmath74 being the correlation length and @xmath75 the linear extension of the @xmath76-dimensional system @xcite . in @xcite
we have developed a microcanonical finite - size scaling theory that takes advantage of the existence of a well - defined transition point @xmath36 in finite microcanonical systems .
the starting point is the scaling behaviour of the entropy of finite systems considered as a function of the energy , the magnetization and the inverse system size .
this then leads to the scaling form @xmath77 of the spontaneous magnetization ( [ eq : msp ] ) . here
@xmath78 is a scaling function characterizing the given universality class and @xmath79 is a nonuniversal constant which is different for the various model systems belonging to the same universality class .
the exponent @xmath80 is given by @xmath81 where @xmath82 is the usual thermal critical exponent governing the divergence of the correlation length on approach to the critical point .
as the spontaneous magnetization presents a square root singularity in any finite microcanonical system , the scaling function @xmath78 varies for small scaling variables @xmath83 as @xmath84 . in figure [ fig_6 ] we test the scaling form ( [ eq : scal ] ) for the usual ising model ( [ eq : ising ] ) with only nearest neighbour interactions ( denoted by i ) and for a generalized ising model with equivalent nearest and next - nearest neighbour interactions ( denoted by gi ) .
the hamiltonian of this latter model is given by @xmath85 where the second sum is over bonds connecting next - nearest neighbours .
the model ( [ eq : gi ] ) belongs to the same universality class as the standard ising model ( [ eq : ising ] ) so that both models should yield the same values for universal quantities .
= 4.0 in figure [ fig_6 ] illustrates two different aspects of the scaling behaviour of the microcanonically defined spontaneous magnetization . on the one hand it shows that both for the ising model and for the gi model a data collapse can be achieved by plotting @xmath86 as a function of @xmath87 .
the values of the involved critical exponents are thereby obtained by the best data collapse .
this yields the values @xmath88 ( @xmath89 ) and @xmath90 ( @xmath91 ) for the ising ( gi ) model , in excellent agreement with the expected values @xmath92 and @xmath93 . on the other hand figure [ fig_6 ]
also proves that the scaling function @xmath78 appearing in eq .
( [ eq : scal ] ) is the same for both models , in accordance with its expected universality .
indeed , the data of both models fall on a common master curve when adjusting the nonuniversal constant @xmath79 , so that @xmath94 . as shown in @xcite scaling functions obtained for models belonging to different universality classes are indeed different . the microcanonical finite - size scaling theory , which has also been applied successfully to the potts @xcite and to the @xmath23 @xcite models , is very promising .
one of the remarkable point is that no _ a priori _ knowledge of the infinite system is needed . especially , one does not need to know the exact location of the critical point of the infinite system .
in this paper we have discussed the signatures of phase transitions in finite microcanonical systems .
discontinuous phase transitions are announced in the microcanonical ensemble by a backbending of the caloric curve and the appearance of negative specific heat .
continuous phase transitions also lead to intriguing finite - size signatures .
indeed , typical features of spontaneous symmetry breaking like the onset of the order parameter or the divergence of the susceptibility are already encountered in finite microcanonical systems .
these singularities are governed by the classical mean - field exponents .
we have discussed in this paper two different approaches that nevertheless enable us to extract the true critical exponents from the density of states of finite systems .
one approach involves effective exponents , whereas the other is based on a microcanonical finite - size scaling theory .
the latter approach explicitly takes advantage of the existence of a well - defined transition point in any finite microcanonical system .
+ + * acknowledgements : * + + it is a pleasure to thank alfred hller for many years of fruitful and enjoyable collaboration .
999 d. h. e. gross , _ microcanonical thermodynamics : phase transitions in small systems _ ( _ lecture notes in physic _ 66 ) ( world scientific , singapore 2001 ) .
w. thirring , z. phys .
* 235 * , 339 ( 1970 ) .
m. bixon and j. jortner , j.chem.phys .
* 91 * , 1631 ( 1989 ) .
a. hller , z. phys .
b * 93 * , 401 ( 1994 ) .
r. s. ellis , k. haven and b. turkington , j. stat . phys . * 101 * , 999 ( 2000 ) .
t. dauxois , p. holdsworth and s. ruffo , eur .
j. b * 16 * , 659 ( 2000 ) .
j. barr , d. mukamel and s. ruffo , phys .
lett . * 87 * , 030601 ( 2001 ) .
i. ispolatov and e. g. d. cohen , physica a * 295 * , 475 ( 2001 ) . f. gulminelli and ph .
chomaz , phys .
e * 66 * , 046108 ( 2002 ) .
m. kastner , m. promberger and a. hller , j. stat . phys . * 99 * , 1251 ( 2000 ) .
d. h. e. gross and e. v. votyakov , eur .
j. b * 15 * , 115 ( 2000 ) .
a. hller and m. pleimling , int .
c * 13 * , 947 ( 2002 ) .
m. pleimling , h. behringer and a. hller , phys .
a * 328 * , 432 ( 2004 ) .
j. hove , phys .
e * 70 * , 056707 ( 2004 ) .
h. behringer , m. pleimling and a. hller , j. phys . a : math .
38 * , 973 ( 2005 ) .
a. richter , m. pleimling and a. hller , cond - mat/0503733 .
h. behringer , j. phys .
gen . * 37 * , 1443 ( 2004 ) .
h. behringer , _ on the structure of the entropy surface of microcanonical systems _ ( mensch und buch verlag , berlin , 2004 ) .
m. dagostino , f. gulminelli , ph .
chomaz , m. bruno , f. cannata , r. bougault , f. gramegna , i. iori , n. le neindre , g. v. margagliotti , a. moroni and g. vannini , phys .
b * 473 * , 219 ( 2000 ) .
m. schmidt , r. kusche , t. hippler , j. donges , w. kronmller , b. von issendorff , and h. haberland , phys .
* 86 * , 1191 ( 2001 ) .
h. behringer , j. phys .
a : math . gen . * 36 * , 8739 ( 2003 ) .
k. binder , physica * 62 * , 508 ( 1972 ) .
d. p. landau , phys .
b * 13 * , 2997 ( 1976 ) .
m.n.barber , in _ phase transitions and critical phenomena _ , vol 8 , edited by c. domb and j. l. lebowitz ( academic press , london and new york , 1983 ) . | the basic quantity for the description of the statistical properties of physical systems is the density of states or equivalently the microcanonical entropy .
macroscopic quantities of a system in equilibrium can be computed directly from the entropy .
response functions such as the susceptibility are for example related to the curvature of the entropy surface .
interestingly , physical quantities in the microcanonical ensemble show characteristic properties of phase transitions already in finite systems . in this paper
we investigate these characteristics for finite ising systems .
the singularities in microcanonical quantities which announce a continuous phase transition in the infinite system are characterised by classical critical exponents .
estimates of the non - classical exponents which emerge only in the thermodynamic limit can nevertheless be obtained by analyzing effective exponents or by applying a microcanonical finite - size scaling theory .
this is explicitly demonstrated for two- and three - dimensional ising systems . |
the anisotropic formulation of lattice qcd has certain advantages when it comes to the study of the equation of state ( eos ) at various finite temperatures .
the finite temperature field theory has a natural asymmetry which makes the anisotropic approach useful to reduce the lattice spacing errors associated with the transfer matrix at less cost than is required for the full continuum limit@xcite . through the introduction of anisotropy on the lattice
one can make the temporal lattice spacing , @xmath0 , sufficiently small so that by varying only the number of time slices , @xmath1 , the temperature can be changed in small discrete steps .
our simulations of full qcd with two flavors of staggered fermions on anisotropic lattices are aimed at the study of the thermodynamic properties of the quark - gluon system .
we employ a fixed parameter scheme in which all the bare parameters of the simulation are kept constant and only the temperature is changed by varying @xmath1 ( from 4 to 64 ) .
this approach separates temperature and lattice spacing effects and keeps the underlying physics scales fixed .
the calculation of the eos of the quark - gluon system involves derivatives of the bare parameters with respect to the physical anisotropy @xmath2 and the spatial lattice spacing @xmath3 .
these `` karsch '' coefficients@xcite are determined as a by - product of the zero - temperature calculations needed to choose the bare parameters .
once determined , the karsch coefficients can be used for all temperatures since they depend only on the intrinsic lattice parameters and not on @xmath1 .
this allows a straight - forward determination of the temperature dependence of the energy and pressure , again at fixed lattice spacing . with two or more slightly different values for @xmath0 , a high - resolution sampling of temperatures can be investigated .
our calculations are based on the qcd action @xmath4 , where the gauge action is : @xmath5 , \nonumber\end{aligned}\ ] ] and the fermion action is : @xmath6 \psi(x)\nonumber\\ & & + \sum_{x } \overline{\psi}(x)\left [ \frac{1}{\xi_0 } \sum_s \slash{d}^{\rm staggered}_s \right ] \psi(x ) .
\nonumber\end{aligned}\ ] ] to derive the eos we start from the thermodynamics identities @xmath7 and @xmath8 using the explicit form for @xmath9 and changing the independent variables with respect to which we differentiate to @xmath10 and @xmath3 we get :
in our simulations we use the r - algorithm@xcite implemented for two flavors of dynamical fermions .
the zero - temperature runs from table 1 are used to calculate the derivatives of the bare parameters with respect to @xmath10 and @xmath3 ( the karsch coefficients ) involved in the eos .
the finite temperature runs in table 2 and 3 represent two separate sweeps through the transition region for each of which the fixed bare parameter scheme described above is applied .
figure 1 shows the variation of @xmath11 through the phase transition as we gradually change the temperature by varying only @xmath1 for @xmath12 and @xmath13 . [ cols="<,<,>,<,<,<",options="header " , ]
figure 2 shows the final result for the eos .
the errors on the pressure are significantly bigger than the errors on the energy due to the large errors on those of the karsch coefficients which are derivatives with respect to @xmath3 .
the comparison with the free lattice theory ( squares ) gives an explanation of the prominent drop off of @xmath14 and @xmath15 in the high temperature sector simply a consequence of the lattice high momentum cut - off .
our results are consistent with previous isotropic results@xcite .
we have studied the qcd thermodynamics using staggered fermions on anisotropic lattices .
the fixed bare parameter scheme allows us to explore the temperature dependence of energy and pressure with fixed physics scales .
while this approach naturally reduces finite lattice spacing errors associated with @xmath0 , the fixed lattice cut - off becomes important at increasing temperature .
including improvements to the spatial parts of the staggered fermion action would be a natural step to reduce the lattice artifacts for high temperatures . 9 qcd - taro collaboration : ph .
de forcrand et al . , _ phys.rev._*d63:054501 * ( 2001 ) f. karsch , _ nucl.phys._*b205[fs5 ] * , 285 ( 1982 ) s. gottlieb et al . , _ phys.rev._*d35 * , 2531 ( 1987 ) milc collaboration : c. w. bernard et al . , _ phys.rev._*d55 * , 6861 ( 1997 ) | numerical simulations of full qcd on anisotropic lattices provide a convenient way to study qcd thermodynamics with fixed physics scales and reduced lattice spacing errors .
we report results from calculations with 2-flavors of dynamical fermions where all bare parameters and hence the physics scales are kept constant while the temperature is changed in small steps by varying only the number of the time slices . the results from a series of zero - temperature scale setting simulations are used to determine the karsch coefficients and the equation of state at finite temperatures . |
the construction of edge ideals associated to ( hyper)graphs ( cf .
@xcite ) provides a viewpoint complementary to the stanley - reisner correspondence in the study of monomial ideals .
edge ideals also provide a framework to study ( hyper)graph theoretic questions from an algebraic perspective .
let @xmath4 be a finite , simple graph over the vertex set @xmath5 .
let @xmath6 be any field and identify the vertices in @xmath7 with the variables in the polynomial ring @xmath8 $ ] .
the edge ideal of @xmath0 is generated by monomials of the form @xmath9 , where @xmath10 is an edge in @xmath0 .
note that an edge can be viewed as a path of length 1 .
thus , for a given positive integer @xmath1 , a more general construction is obtained by considering monomials corresponding to paths of length @xmath3 in @xmath0 .
this is the _ path ideal _ construction .
path ideals were first introduced by conca and de negri in @xcite , and their algebraic properties have been investigated by various authors in the literature ( cf .
@xcite ) . in this paper , we shall study path ideals of _ rooted trees_. recall that a _ tree _ is a graph in which there exists a unique path between every pair of distinct vertices ; a _ rooted tree _ is a tree together with a fixed vertex called the _
root_. in particular , in a rooted tree there exists a unique path from the root to any given vertex .
we can also view a rooted tree as a _ directed _ graph by assigning to each edge the direction that goes `` away '' from the root .
_ throughout this paper , a rooted tree will always be viewed as a directed , rooted tree in this sense_. if @xmath11 is an edge in a rooted tree @xmath0 , then we write @xmath12 for the `` directed '' edge whose direction is from @xmath13 to @xmath14 .
the path ideal of a rooted tree is defined in precise form as follows .
let @xmath15 be a given integer , and let @xmath0 be a rooted tree with vertex set @xmath16 . 1 .
a directed _ path _ of length @xmath3 is a sequence of distinct vertices @xmath17 , in which @xmath18 is the directed edge from @xmath19 to @xmath20 for any @xmath21 .
the _ path ideal _ of length @xmath3 associated to @xmath0 is the monomial ideal @xmath22.\ ] ] in particular , when @xmath23 , @xmath24 is the maximal homogeneous ideal , which is well understood .
hence , all of our results in this paper will be for path ideals of length at least 1 ( i.e. , @xmath25 ) .
due to the correspondence between paths and monomials we shall often abuse notation and use @xmath26 to denote both the monomial @xmath26 in @xmath27 $ ] and the path @xmath17 in @xmath0 .
a _ rooted forest _ is a disjoint union of rooted trees . for a rooted forest @xmath28
, we define the path ideal @xmath29 to be the sum of the path ideals of the connected components of @xmath28 .
a path ideal @xmath2 is a squarefree monomial ideal , so it can also be realized as the edge ideal of a hypergraph or the stanley - reisner ideal of a simplicial complex .
given a path ideal , the corresponding hypergraph and simplicial complex are in general very complicated .
the goal of this paper is then to investigate algebraic properties and invariants of a path ideal @xmath2 via the combinatorial structures of the rooted tree @xmath0 .
we are interested in invariants associated to the minimal free resolution of @xmath2 , namely , the graded betti numbers , the castelnuovo - mumford regularity , and the projective dimension .
we now provide an overview of the structure of the paper and our results . in section [ treesection ] ,
we recall some useful notation and terminology , and prove our first main result ; here we give a recursive formula to compute the graded betti numbers of path ideals ( theorem [ nocancellation ] and remark [ rmk.recursive ] ) .
section [ sec.regularity ] is devoted to studying the castelnuovo - mumford regularity of path ideals .
the main result of this section , theorem [ thm.reg ] , provides a general bound for the regularity of a path ideal in terms of the number of leaves and the number of pairwise disjoint paths of length @xmath1 in the tree . in section [ sec.linear ]
, we study the linear strand of @xmath2 and classify all rooted trees @xmath0 for which @xmath2 has a linear resolution .
our first result of this section , theorem [ thm.linearstrand ] , gives a precise formula for graded betti numbers @xmath30 on the linear strand of @xmath2 .
our next result in this section , theorem [ thm.ntp ] , shows that @xmath2 has a linear resolution if and only if it has linear first syzygies ; this is the case if any only if @xmath0 belongs to a special class of rooted trees , which we will call broom graphs . in section [ line_graphs ] ,
we restrict our attention to rooted trees occurring as path graphs . for a path graph @xmath0 , in theorem [ thm.nonzero ] , we characterize which graded betti numbers of @xmath2 are nonzero . as a consequence , we compute the regularity of @xmath31 explicitly in corollary [ thm.reglinegraph ] .
we also recover he and van tuyl s formula for the projective dimension of @xmath31 in this case ( corollary [ pdlinegraph ] )
. * acknowledgement .
* this project started when the first author visited the other authors at tulane university .
the authors wish to thank tulane university for its hospitality .
the authors would also like to thank adam van tuyl for stimulating discussions and suggestions , and to thank the two anonymous referees for many useful comments making the paper more readable .
the second author acknowledges support from the board of regents grant leqsf(2007 - 10)-rd - a-30 .
from this point forward , @xmath0 will denote a rooted tree ( also viewed as a directed tree ) with vertex set @xmath5 , @xmath6 will denote a field of arbitrary characteristic , and @xmath1 will denote a given positive integer . then @xmath8 $ ] will denote the corresponding polynomial ring .
* induced subgraphs and examples .
* we will now introduce some combinatorial terminology and provide examples of path ideals .
let @xmath0 be a rooted tree with root @xmath32 .
for a given vertex @xmath33 in @xmath0 , the _ level _ of @xmath33 , denoted @xmath34 , is defined to be the length of the unique path from @xmath32 to @xmath33 .
the _ height _ of @xmath0 , denoted @xmath35 , is the maximal level of vertices in @xmath0 .
sometimes we will need to consider rooted forests .
the level of a vertex @xmath33 in a rooted forest @xmath28 is defined to be the level of @xmath33 inside the connected component of @xmath28 containing @xmath33 .
the height of a rooted forest @xmath28 is defined to be the largest height of its connected components .
consider the following rooted tree @xmath0 .
the edges in @xmath0 are given directions that go `` away '' from the root , making @xmath0 a directed tree .
for instance , there is a unique path @xmath36 going from the root @xmath37 to the vertex @xmath38 , and a unique path @xmath39 going from the root @xmath37 to the vertex @xmath40 ; and so , the direction of the edge @xmath41 is from @xmath38 to @xmath40 .
it can also be seen that the highest level in @xmath0 is 4 ( @xmath42 ) , so @xmath43 . for @xmath44 , and @xmath45
, we have the following path ideals associated to @xmath0 : @xmath46 notice that the path ideals of @xmath0 depend on the choice of the root of @xmath0 .
let @xmath0 be a rooted tree , and let @xmath32 be a vertex in @xmath0 . 1 .
a vertex @xmath47 in @xmath0 is the _ parent _ of @xmath32 if and only if @xmath48 is a directed edge in @xmath0 .
a vertex @xmath33 is called a _ child _ of @xmath32 if @xmath49 is a directed edge in @xmath0 .
a vertex @xmath50 is an _ ancestor _ of @xmath32 if there is a path from @xmath47 to @xmath32 .
a vertex @xmath51 is a _ descendant _ of @xmath32 if there is a path from @xmath32 to @xmath33 .
the vertex @xmath32 is called a _ leaf _ of @xmath0 if @xmath32 has no descendants .
4 . the vertex @xmath32 is called the _ root _ of @xmath0 if @xmath32 has no ancestors . 5 .
the _ degree _ of a vertex @xmath32 in @xmath0 , denoted by @xmath52 , is the number of edges in @xmath0 incident to @xmath32 . 1 .
let @xmath53 be a finite simple graph .
a subgraph @xmath54 of @xmath53 is called an _ induced subgraph _ if for every pair of vertices @xmath55 in @xmath54 the following condition holds : if @xmath56 is an edge in @xmath53 , then it is also an edge in @xmath54 .
2 . let @xmath0 be a rooted tree .
an _ induced subtree _ ( or _ forest _ ) of @xmath0 is a directed subtree ( or forest ) that is also an induced subgraph of @xmath0 .
3 . let @xmath0 be a rooted tree and let @xmath32 be a vertex in @xmath0 .
the _ induced subtree rooted at @xmath32 _ of @xmath0 is the induced subtree of @xmath0 over the vertex set @xmath57 ( with @xmath32 considered as its root )
. * notation .
* let @xmath0 be a rooted tree , and let @xmath58 be a collection of vertices in @xmath0 .
we shall denote by @xmath59 the induced subforest of @xmath0 obtained by removing the vertices in @xmath58 and the edges incident to these vertices .
if @xmath58 consists of a single element @xmath32 , then we write @xmath60 for @xmath61 .
* minimal free resolutions .
* let @xmath8 $ ] , and let @xmath62 be a finitely generated graded @xmath63-module .
associated to @xmath62 is a _ minimal free resolution _ , which is a finite complex of the form @xmath64 where the maps @xmath65 are exact and where @xmath66 denotes the translation of @xmath63 obtained by shifting the degrees of elements of @xmath63 by @xmath67 .
the numbers @xmath68 are called the _ graded betti numbers _ of @xmath62 , and they provide the number of minimal generators of degree @xmath67 occuring in the @xmath69th - syzygy module of @xmath62 .
if @xmath62 is generated in degree @xmath1 , then the _ linear strand _ of @xmath62 is given by the betti numbers @xmath70 , for @xmath71 .
in this case , @xmath62 is said to have linear first syzygies if @xmath72 for all @xmath73 ; and more generally , @xmath62 is said to have a _ linear resolution _ if @xmath74 for all @xmath71 and @xmath75 .
we are interested in the following two invariants that measure the `` size '' of the minimal free resolution .
let @xmath63 and @xmath62 be as above . 1 .
the _ projective dimension _ of @xmath62 , denoted by @xmath76 , is the length of the minimal free resolution associated to @xmath62 .
2 . the _ castelnuovo - mumford regularity _ ( or simply , _
regularity _ ) , denoted by @xmath77 , is a measure of the width of the minimal free resolution of @xmath62 and is defined as @xmath78 [ lem.ringextension ] let @xmath8 $ ] , and let @xmath62 be a graded @xmath63-module .
let @xmath79 be indeterminates , and denote by @xmath80 the polynomial ring @xmath81 $ ] .
then @xmath82 where the second regularity is computed for the @xmath80-module @xmath83 .
it is clear that the ring extension @xmath84 is flat .
thus , tensoring with @xmath80 maps a minimal free resolution to a minimal free resolution .
the result now follows from the definition of regularity .
lemma [ lem.ringextension ] allows us to look at extensions of ideals in rings with more variables when discussing regularity .
for instance , if @xmath0 is a rooted tree corresponding to the polynomial ring @xmath80 and @xmath28 is an induced , rooted subtree of @xmath0 corresponding to the polynomial ring @xmath63
( i.e. the variables in @xmath80 correspond to the vertices in @xmath0 , and the variables in @xmath63 correspond to the vertices in @xmath28 ) , then we can abuse notation and write @xmath29 for both the path ideal of @xmath28 in @xmath63 and also for its extension in the bigger ring @xmath80 when discussing regularity .
* mapping cone decomposition of path ideals . *
to study the minimal free resolutions of the quotient rings @xmath31 , we provide an inductive construction of the path ideals via the mapping cone construction .
this construction is a generalization of the method provided for edge ideals in @xcite .
this method will allow the decomposition of a given path ideal into a collection of simpler path ideals corresponding to smaller trees . given a short exact sequence @xmath85 where @xmath86 , @xmath87 , and @xmath88 are graded @xmath63-modules ,
the mapping cone is a method to construct a free resolution for @xmath88 knowing free resolutions of @xmath86 and @xmath87 ( for more details on the mapping cone construction we refer the reader to @xcite ) . in general , given _
minimal _ free resolutions for @xmath86 and @xmath87 , the mapping cone construction does not necessarily give a _
minimal _ free resolution of @xmath88 .
however , in the case of path ideals , we shall show that the mapping cone construction does indeed provide a minimal free resolution for a particular short exact sequence .
[ nocancellation ] let @xmath0 be a rooted tree with vertex set @xmath16 and height @xmath89 .
let @xmath90 denote a leaf of @xmath0 of level at least @xmath3 .
then by letting @xmath91 denote the path terminating at @xmath90 , the mapping cone procedure applied to the sequence @xmath92 provides a minimal free resolution of @xmath31 .
in particular , for any @xmath69 and @xmath67 , we have @xmath93 since @xmath90 does not divide a minimal generator of @xmath94 , @xmath95 however , this implies that the exact sequence @xmath96 factors as @xmath97 & \big(s\big / i_t(\gamma \backslash x_{i_{t } } ) : ( x_{i_1}\cdots x_{i_t})\big)(-t ) \ar[r]^{\mbox{\hspace{.6in}}x_{i_1}\cdots x_{i_t}}\ar[dd]^{x_{i_t } } & s / i_t(\gamma \backslash x_{i_{t } } ) \ar[r ] & s / i_t(\gamma ) \ar[r ] & 0.\\ & & & & \\ & \big(s\big / i_t(\gamma \backslash x_{i_{t } } ) : ( ( x_{i_1}\cdots x_{i_{t-1}}))\big)(-t+1 ) \ar[uur]^{x_{i_1}\cdots x_{i_{t-1 } } } & & & } \label{eq : factor } \end{aligned}\ ] ] let @xmath98 be minimal free resolutions of @xmath99 and @xmath100 respectively .
then the mapping cone construction applied to the short exact sequence provides a free resolution of @xmath31 given by @xmath101 where the maps @xmath102 are defined by @xmath103 $ ] and @xmath104\mbox{\hspace{.25in}for } i>1 , \label{eq : sigma } \end{aligned}\ ] ] ( @xmath105 are resulted from the homomorphism @xmath106 ) . from the factorization given in , the entries of the matrix of @xmath65 are not units .
furthermore , since and are minimal free resolutions , the matrix representation of @xmath102 in can not contain units .
therefore , the mapping cone construction applied to provides a minimal free resolution of @xmath31 .
in particular , this implies that @xmath107 for all @xmath69 and @xmath67 .
theorem [ nocancellation ] provides an inductive method to study algebraic properties of @xmath2 as the colon ideal @xmath108 can be realized as a disjoint union of path ideals of varying lengths .
[ lem.subtree ] let @xmath0 be a rooted tree of height @xmath89 , let @xmath90 be a leaf at the highest level in @xmath0 , and let @xmath17 be the unique path of length @xmath3 terminating at @xmath90 .
let @xmath109 be the only parent of @xmath110 , if it exists .
for @xmath111 , let @xmath112 be the induced subtree of @xmath0 rooted at @xmath19 , and let @xmath113 for @xmath114 .
then @xmath115 let @xmath53 be the set of minimal generators of @xmath116 , i.e. elements in @xmath53 corresponding to paths of length @xmath3 in @xmath117 .
clearly , @xmath118 observe first that @xmath119 is a path of length @xmath3 in @xmath117 , and @xmath120 .
consider a path @xmath121 of length @xmath3 in @xmath117 that does not contain @xmath109 .
there are three possibilities for @xmath121 .
* case 1 : * @xmath121 contains none of the vertices in @xmath122 , and @xmath121 is not a path in the induced subtree rooted at @xmath109 .
this is the case if and only if @xmath123 . * case 2 : * @xmath121 contains none of the vertices in @xmath122 , and @xmath121 is a path in the induced subtree rooted at @xmath109 .
this is the case if and only if @xmath121 is a path of length @xmath3 in the rooted forest @xmath124 .
* case 3 : * @xmath121 contains some but not all of the vertices @xmath125 , and @xmath121 does not contain @xmath109 . let @xmath126 be the largest index such that @xmath127 is in @xmath121 . since @xmath90 is a leaf of highest level in @xmath0 , @xmath121 can contain at most @xmath128 descendants of @xmath127 .
this implies that @xmath121 must contain all the vertices @xmath129 .
this is the case only if @xmath130 is a path of length @xmath131 in @xmath132 .
furthermore , because @xmath90 is of highest level in @xmath0 , any path of length @xmath133 in @xmath134 must be from a child of @xmath127 ( other than @xmath135 ) to a leaf in @xmath134 .
thus , case 3 appears if and only if @xmath136 [ rmk.recursive ] note that in lemma [ lem.subtree ] , @xmath137 ( or @xmath138 if @xmath109 does not exist ) since @xmath90 is a leaf at the highest level in @xmath0 .
observe further that the minimal generators of the ideals @xmath139 , and @xmath140 involve pairwise disjoint sets of vertices .
thus , the minimal free resolution of @xmath141 $ ] is obtained by taking the tensor product of the minimal free resolutions of @xmath142 , @xmath143 , and @xmath144 for @xmath145 .
together with theorem [ nocancellation ] , this gives a recursive formula to compute the graded betti numbers of @xmath2 .
in particular , the graded betti numbers of @xmath2 do not depend on the characteristic of the ground field @xmath6 .
this fact was proved in ( * ? ? ?
* theorem 3.1 ) .
it is also a corollary of a more general recursive formula for the graded betti numbers of simplicial forests given in ( * ? ? ?
* theorem 5.8 ) .
in this section , we give a bound for the regularity of @xmath2 . from the alexander duality ( cf .
* theorem 5.59 ) ) , one obtains the following trivial bound @xmath146 , where @xmath147 is the alexander dual of @xmath2 .
we are seeking a bound for @xmath148 that is , in general , better than @xmath149 .
our bound will be based on the number of leaves and the number of pairwise disjoint paths of length @xmath1 in @xmath0 .
let @xmath0 be a rooted tree .
we define @xmath150 to be the number of leaves in @xmath0 whose level is at least @xmath151 and @xmath152 to be the maximal number of pairwise disjoint paths of length @xmath1 in @xmath0 ( i.e. , @xmath153 ) .
note that , in general , @xmath154 and @xmath155 . in the next few corollaries , @xmath0 will denote
a rooted tree of height @xmath89 , and @xmath90 will denote a leaf of highest level in @xmath0 .
let @xmath17 be the unique path of length @xmath3 terminating at @xmath90 , and let @xmath109 be the parent of @xmath110 ( if it exists )
. set @xmath156 ( or @xmath157 if @xmath109 does not exist ) .
furthermore for @xmath145 , let @xmath112 be the induced subtree of @xmath0 rooted at @xmath19 , and let @xmath113 .
[ cor.reg ] we have @xmath158 it is easy to see that for @xmath145 the minimal generators of the ideals @xmath159 , @xmath160 , and @xmath161 involve pairwise disjoint sets of vertices .
thus , by lemma [ lem.subtree ] , the minimal free resolution of @xmath162 is given by the tensor product of the minimal free resolution of @xmath163 , @xmath143 , and @xmath164 for @xmath145 .
this implies that @xmath165 the conclusion now follows from the fact that @xmath166 [ cor.smalltree ] we have @xmath167 in particular , by considering @xmath168 in place of @xmath0 we have @xmath169 if @xmath109 does not exist , then @xmath170 it follows from theorem [ nocancellation ] that @xmath171 the first conclusion follows by applying corollary [ cor.reg ] .
the second conclusion follows by observing that since @xmath90 is a leaf at the highest level , we have @xmath172 ( or @xmath173 if @xmath109 does not exist ) .
we are ready to prove our next theorem .
[ thm.reg ] let @xmath0 be a rooted tree over the vertex set @xmath5
. then @xmath174.\ ] ] we shall use induction on both @xmath1 and @xmath175 . for @xmath23 ,
the ideal @xmath2 is the maximal homogeneous ideal of @xmath8 $ ] , and the assertion is clearly true .
assume that @xmath25 .
the assertion is also true if @xmath176 , so we may assume that @xmath177 .
let @xmath178 .
observe that if @xmath179 then @xmath180 , making the assertion vacuous .
we shall assume that @xmath89 .
consider first the case when @xmath181 .
in this case , any path of length @xmath3 in @xmath0 must be from the root to a leaf ( at level @xmath3 ) of @xmath0 and so @xmath182 . without loss of generality , assume that @xmath37 is the root of @xmath0 . then @xmath183 and @xmath184 . by the induction hypothesis , we have @xmath185 = ( t-2)l_{t-1}(\gamma \backslash x_1).\ ] ] observe further that @xmath186 ( since @xmath181 , @xmath0 must have at least a leaf at level @xmath3 ) . therefore , we have @xmath187 and the assertion is true .
consider now the case when @xmath188 .
let @xmath189 be a leaf at the highest level , and let @xmath190 be the unique path of length @xmath1 terminating at @xmath90 .
let @xmath156 , let @xmath112 be the induced subtree of @xmath0 rooted at @xmath19 , and let @xmath113 for @xmath145 .
it follows from corollary [ cor.smalltree ] that @xmath191 observe that @xmath192 and @xmath193 .
thus , by the induction hypothesis , we have @xmath194 \le ( t-1)[l_t(\gamma ) + p_t(\gamma)].\ ] ] it can also be seen that @xmath195 and @xmath196 .
thus , by the induction hypothesis , we have @xmath197 + ( t-1 ) \\ & = ( t-1 ) l_t(\gamma ) + ( t-1)[p_t(\gamma \backslash \gamma_0 ) \\ & \quad + \sum_{j=0}^{t-1 } p_{t - j}(\delta_j \backslash p ) ] + ( t-1).\end{aligned}\ ] ] moreover , @xmath198 , @xmath199 , and @xmath200 ( because @xmath201 for any @xmath67 ) . hence , @xmath202 + ( t-1 ) \\ & \le ( t-1)[l_t(\gamma ) + p_t(\gamma)].\end{aligned}\ ] ] the theorem is proved by the use of .
the bound in theorem [ thm.reg ] is sharp when @xmath0 is a disjoint union of paths of length @xmath3 .
for instance , if @xmath0 is a directed path @xmath203 of length @xmath3 , then @xmath204 and @xmath182 .
hence , @xmath205.$ ]
in this section , we compute the linear strand of @xmath2 for a rooted tree @xmath0 , and classify all rooted trees @xmath0 for which @xmath2 has a linear resolution .
we start by investigating the linear strand of @xmath2 .
note that @xmath206 is just the number of paths of length @xmath3 in @xmath0
. therefore we will be interested in @xmath30 for @xmath207 .
note also that the path ideal @xmath2 can be realized as the edge ideal of a _ hyper - tree_. in @xcite , the second author and van tuyl gave a formula for the linear strand of the edge ideal of any hyper - tree .
however , the structure of the hyper - tree corresponding to @xmath2 is quite complicated .
we shall use the combinatorial data of @xmath0 to provide an explicit formula for the linear strand of @xmath2 .
[ lem.ls1 ] let @xmath0 be a rooted tree of height @xmath89 , let @xmath90 be a leaf at the highest level in @xmath0 , and let @xmath17 be the unique path of length @xmath3 terminating at @xmath90 .
then for @xmath71 , @xmath208 let @xmath109 be the parent of @xmath110 if it exists ; and let @xmath209 .
furthermore , for @xmath145 , let @xmath112 be the induced subtree of @xmath0 rooted at @xmath19 , and let @xmath113 .
observe , as before , that the minimal generators of the ideals @xmath159 , @xmath160 , and @xmath161 involve pairwise disjoint sets of vertices .
thus , by lemma [ lem.subtree ] , the minimal free resolution of @xmath210 is the tensor product of the minimal free resolutions of @xmath163 , @xmath143 , and @xmath164 for @xmath211 .
therefore , the contribution to @xmath212 comes from @xmath213 .
observe that the number of vertices of @xmath214 is @xmath215 if @xmath216 or @xmath217 and @xmath218 ( i.e. , when @xmath219 exists ) .
if @xmath220 and @xmath221 ( i.e. , when @xmath219 does not exist ) then the number of vertices of @xmath214 is @xmath222 .
observe further that @xmath109 exists if @xmath218 .
hence , the conclusion follows from the fact that the minimal free resolution of @xmath223 is the koszul complex .
[ thm.linearstrand ] let @xmath0 be a rooted tree over the vertex set @xmath7 .
then for @xmath207 , @xmath224 let @xmath225 .
the assertion is vacuously true
if @xmath179 so we may assume that @xmath89 .
we shall use induction on @xmath175 , the number of vertices in @xmath0 .
again , the assertion is vacuously true if @xmath226 .
assume that @xmath177 .
let @xmath90 be a leaf at the highest level in @xmath0 , and let @xmath17 be the unique path of length @xmath3 terminating at @xmath90 . for simplicity ,
let @xmath227 . as observed before , @xmath228 . by theorem [ nocancellation ] , we have @xmath229 if @xmath217
then by the induction hypothesis and lemma [ lem.ls1 ] , we have @xmath230 since @xmath231 for all @xmath232 , @xmath233 , and @xmath234 , we have @xmath235 here , the last equality follows by adding @xmath236 .
assume now that @xmath216 .
consider the case when @xmath181 . by the induction hypothesis and lemma [ lem.ls1 ] , we have @xmath237 since @xmath181 and @xmath90 is at the highest level
, we have @xmath238 . also , @xmath231 for all @xmath239 and @xmath240 .
thus , @xmath241 as before , we can add @xmath236 to the sum and the assertion follows .
now we consider the case @xmath218 . by the induction hypothesis and lemma [ lem.ls1 ] ,
we have @xmath242 since @xmath218 and @xmath243 , we have @xmath244 . as before , @xmath231 for all @xmath239 and @xmath240 .
thus , using a similar computation as above , we get @xmath245 the rest of the section is devoted to classifying all rooted trees @xmath0 for which @xmath2 has a linear resolution .
recall that @xmath2 has a linear resolution if and only if @xmath246 for all @xmath75 .
our characterization is based on the following special class of rooted trees .
[ defn.broom ] a _ broom graph _ of type @xmath1 consists of a _ handle _ , which is a directed path @xmath247 , such that every edge in @xmath0 ( not on the handle ) has the form @xmath248 for some @xmath249 .
since @xmath1 is usually fixed in our context , we often omit the phrase `` of type @xmath1 '' and refer to a broom graph of type @xmath1 simply as a broom graph .
note also that in a broom graph as defined @xmath126 is related to the height of the graph .
if @xmath32 is a leaf of level strictly less than @xmath3 , then @xmath250 and @xmath2 have the same generators ( in different polynomial rings ) .
this implies that @xmath251 for all @xmath252 .
thus we can successively remove leaves at level strictly less than @xmath3 from a rooted tree without changing the graded betti numbers of its path ideal .
we call this process the _ cleaning _ process of @xmath0 . the rooted tree obtained after the cleaning process
is called the _ clean form _ of @xmath0 , denoted @xmath253 . we are now ready to state our characterization . [ thm.ntp ]
let @xmath0 be a rooted tree of height @xmath89 .
then the following are equivalent : 1 .
@xmath2 has linear first syzygies .
2 . @xmath2 has a linear resolution .
@xmath253 is a broom graph of height at most @xmath254 . without loss of generality
, we may assume that @xmath0 is already in its clean form , ie .
, @xmath255 . in this case , all leaves of @xmath0 have level at least @xmath3 .
we shall use induction on @xmath175 , the number of vertices in @xmath0 .
the statement is clearly true for @xmath176 ( since in this case @xmath2 is either @xmath256 or @xmath257 ) .
assume that @xmath177 .
as before , let @xmath90 be a leaf at the highest level in @xmath0 , and let @xmath17 be the unique path of length @xmath3 terminating at @xmath90 .
let @xmath109 be the parent of @xmath110 if it exists ; and let @xmath156 .
furthermore , for @xmath145 , let @xmath112 be the induced subtree of @xmath0 rooted at @xmath19 , let @xmath113 , and let @xmath227 .
by theorem [ nocancellation ] , we have @xmath258 this implies that @xmath2 has linear first syzygies ( resp .
, has a linear resolution ) if and only if @xmath259 has linear first syzygies ( respectively , has a linear resolution ) , and @xmath260 is generated in degree one ( respectively , is generated in degree one and has a linear resolution ) . by lemma [ lem.subtree ]
the minimal free resolution of @xmath261 is the tensor product of the minimal free resolutions of @xmath163 , @xmath143 , and @xmath164 .
it follows that @xmath260 is generated in degree one if and only if @xmath262 and @xmath263 for all @xmath264 .
note that if this is the case then @xmath260 has a linear resolution .
thus , ( 1 ) and ( 2 ) are equivalent .
now we show that ( 2 ) @xmath265 ( 3 ) . since @xmath0 is in its clean form ,
if @xmath262 then @xmath266 must be a directed path from the root of @xmath0 to the parent of @xmath109 of length strictly less than @xmath3 . to show that @xmath0 is a broom , it now suffices to show that for any @xmath67 , @xmath267 consists of isolated vertices .
suppose for some @xmath264 , @xmath267 contains a path of length at least 1 . in this case
, we can find a path of length at least 2 of the form @xmath268 in @xmath269 terminating at a leaf @xmath270 for some @xmath271 . here , the second inequality is due to the fact that @xmath90 is of highest level .
recall that for @xmath2 to have a linear resolution , @xmath272 must also have a linear resolution .
thus by induction and successively removing vertices , we will reduce @xmath0 to a rooted tree @xmath273 in which @xmath274 is a leaf at the highest level and @xmath275 has a linear resolution .
let @xmath276 be the graph rooted at @xmath19 obtained from @xmath273 in the same fashion as how @xmath269 was obtained from @xmath0 ( with @xmath274 replacing the role of @xmath90 ) .
let @xmath277 be the set of vertices on the unique path of length @xmath3 terminating at @xmath274 ( this path exists since the level of @xmath270 is at least @xmath3 ) . by a similar argument as with @xmath269 , for @xmath275 to have a linear resolution we must have @xmath278
. however , this is not true since @xmath279 .
we have now shown that for each @xmath264 , @xmath267 consists of isolated vertices .
since @xmath90 is of highest level , @xmath214 also consists of isolated vertices .
we can conclude that if @xmath2 has a linear resolution , then @xmath0 is a broom of height at most @xmath254 .
conversely , suppose that @xmath0 is a broom of height at most @xmath254 in its clean form . by definition , it is easy to see that in this case , @xmath59 consists of a path of length at most @xmath280 along with isolated vertices , and @xmath267 consists of isolated vertices for any @xmath67 .
thus , @xmath262 , @xmath263 for all @xmath264 , and @xmath281 has a linear resolution .
moreover , since @xmath117 is also a broom of height at most @xmath254 , by the induction hypothesis , @xmath116 has a linear resolution .
thus , @xmath2 has a linear resolution
. therefore , ( 3 ) @xmath265 ( 2 ) . for any rooted tree @xmath0 , a cellular complex supporting the linear strand of @xmath2
is described by the minimal generators of @xmath2 ; specifically , if @xmath282 are minimal generators of @xmath2 , then @xmath283 form a cell if and only if it is maximal with respect to the property that @xmath284 when @xmath0 is a broom graph of height at most @xmath254 , this cellular complex also supports the minimal free resolution of @xmath2 . in this case ,
set @xmath285 and let @xmath286 be a vertex of degree @xmath287 with highest level .
then it follows from theorem [ thm.linearstrand ] that this cellular complex supporting the minimal free resolution of @xmath2 has dimension ( equivalently , the projective dimension of @xmath2 ) equal to @xmath288
in this last section of the paper , we restrict our attention to a simple class of rooted trees , namely path graphs .
the _ path graph _ over the vertex set @xmath5 is the directed tree whose directed edges ( after a possible re - indexing ) are @xmath289 for @xmath290 .
we shall first show that ( 1 ) @xmath265 ( 2 ) . to have @xmath311 , clearly @xmath312 .
it suffices to show that @xmath313 for some @xmath126 satisfying @xmath314 . by the recursive formula ( [ eq.linegraph ] ) , we have that at least one of the graded betti numbers @xmath315 , @xmath316 or @xmath317 must be nonzero . if @xmath318 then , by the induction hypothesis , we have @xmath313 where @xmath319 and @xmath320 .
and so ( 2 ) follows .
if @xmath321 , then by the induction hypothesis , we have @xmath322 for some @xmath323 and @xmath324 . by taking @xmath325 ,
( 2 ) again follows . in the case where @xmath326 , a similar argument again implies that once again ( 2 ) follows . if @xmath331 ( i.e. , @xmath332 ) , then set @xmath333 . by remark [ rmk.linegraph ]
, we have @xmath334 .
since @xmath335 and @xmath336 , the induction hypothesis now implies that @xmath326 , and ( 1 ) follows from .
it remains to consider the case where @xmath337 .
this , in particular , implies that @xmath338 .
observe that if @xmath339 , @xmath340 implies that @xmath341 . on the other hand , if @xmath342 for some @xmath343 , then @xmath344 and we have @xmath345 .
therefore we have @xmath346 ; that is , @xmath347 .
also , @xmath348 .
now , set @xmath333 , and observe that @xmath349 and @xmath350 .
the induction hypothesis implies that @xmath351 , and ( 1 ) follows again from .
99 rachelle bouchat , _ free resolutions of some edge ideals of simple graphs _ , journal of commutative algebra * 2 * ( 2010 ) , no.1 , 1 - 36 .
paulo brumatti and aparecida francisco da silva , _ on the symmetric and rees algebras of @xmath362-cyclic ideals _ ,
* 21 * ( 2001 ) , 27 - 42 , 16th school of algebra , part ii ( portuguese ) , braslia , 2000 .
aldo conca and emanuela de negri , _ @xmath62-sequences , graph ideals , and ladder ideals of linear type _ , journal of algebra * 211 * ( 1999 ) , no.2 , 599 - 624 .
huy ti h and adam van tuyl , _ splittable ideals and the resolutions of monomial ideals _ , journal of algebra * 309 * ( 2007 ) , no . 1 , 405 - 425 . huy ti h and adam van tuyl , _ monomial ideals , edge ideals of hypergraphs , and their graded betti numbers _ , journal of algebraic combinatorics * 27 * ( 2008 ) , no . 2 , 215 - 145 .
jing ( jane ) he and adam van tuyl , _ algebraic properties of the path ideal of a tree _
, arxiv:0902.0902 .
gaetana restuccia and rafael h. villarreal , _ on the normality of monomial ideals of mixed products _ ,
algebra * 29 * ( 2001 ) , no . 8 , 3571 - 3580 .
ezra miller and bernd sturmfels , _ combinatorial commutative algebra _
, springer gtm 227 , springer , 2004 .
rafael h. villarreal , _
rees algebras of edge ideals _ , comm .
algebra * 23 * ( 1995 ) , no . 9 , 3513 - 3524 .
charles a. weibel , _ an introduction to homological algebra _ , cambridge studies in advanced mathematics 38
. cambridge university press , cambridge , 1994 . | let @xmath0 be a rooted ( and directed ) tree , and let @xmath1 be a positive integer .
the path ideal @xmath2 is generated by monomials that correspond to directed paths of length @xmath3 in @xmath0 . in this paper
, we study algebraic properties and invariants of @xmath2 .
we give a recursive formula to compute the graded betti numbers of @xmath2 in terms of path ideals of subtrees .
we also give a general bound for the regularity , explicitly compute the linear strand , and investigate when @xmath2 has a linear resolution . |
for a long time , the anomalous hall effect ( ahe ) attracts increasing attraction from both theoretical and experimental viewpoint . in general , the hall resistivity is given by @xmath6 , where @xmath7 is the magnetic field and @xmath8 is the ordinary hall coefficient due to the lorentz force .
@xmath9 is the anomalous hall coefficient in the presence of magnetization @xmath10 . in ferromagnets , @xmath9 is usually at least one order of magnitude larger than @xmath8 . in paramagnetic heavy fermion ( hf ) systems , @xmath11 takes a huge value due to the ahe since the uniform susceptibility @xmath12 in hf is about @xmath13 times larger than that in usual metals owing to the strong coulomb interaction .
recently , the ahe in transition metal ferromagnets has been intensively studied experimentally .
for example , the ahe in ferromagnetic pyrochlore oxides shows interesting behaviors @xcite .
the ahe is also observed in ferromagnetic spinels @xcite and in ru , ti oxides @xcite .
several multilayer systems @xcite also show distinct ahe .
the ahe in spin glass systems had been discussed in refs .
@xcite .
the theoretical study of ahe was initiated by karplus and luttinger ( kl ) @xcite in 1954 .
they found that the anomalous hall conductivity ( ahc ) @xmath14 is finite and dissipation - less ( i.e. , @xmath15 is independent of resistivity @xmath2 ) when @xmath16 .
this kl - term is called the `` intrinsic ahe '' because it exists independent of impurities . in 1958 , smit presented a mechanism of `` extrinsic ahe '' @xcite : he found that spin polarized electrons are scattered asymmetrically around an impurity in the presence of spin - orbit coupling .
the ahc due to this skew - scattering mechanism is linearly proportional to @xmath2 if elastic scattering is dominant .
the above two works were reproduced and refined in a unified way by luttinger @xcite by using the transport theory of kohn and luttinger , and his work were also reproduced by linear - response theory using diagrammatic technique @xcite . in 1970 , berger proposed another mechanism of extrinsic ahe , the side jump due to impurities @xcite .
this mechanism gives the ahc in proportion to @xmath17 .
note that the extrinsic ahe vanishes in a clean system without impurity . for a long time
, the ahe had been generally regarded as an extrinsic effect , and the intrinsic ahe proposed by kl had been underestimated .
this would be because the theoretical model assumed in ref .
@xcite was too oversimplified .
moreover , ref .
@xcite could not offer a specific expression for @xmath18 .
recently , however , various experiments suggest that the intrinsic ahe @xmath19 is dominant in many transition metal ferromagnets @xcite even in good metals with @xmath20 : it is surprising because the extrinsic ahc @xmath21 should be dominant when @xmath2 is small , if inelastic scattering is negligibly small .
we here remind readers that the extrinsic ahc is very sensitive to the statistical property of impurity potential @xmath22 ; for instance , it vanishes when the average value of @xmath23 is zero @xcite .
after luttinger s work @xcite , theory of intrinsic ahe has been developed by many authors . in general , @xmath24 is composed of the `` fermi surface term '' and the `` fermi sea term '' ; the latter comes from quasiparticles inside of the fermi sea , and it could exist even in insulators @xcite .
( note that the conductivity @xmath25 is composed of only the fermi surface term . )
the term of the ahc derived by kl is a part of the `` fermi sea term '' .
recently , m. onoda and nagaosa @xcite and sundaram and niu @xcite found that kl s ahc is expressed in terms of the `` berry curvature '' @xcite .
( note that luttinger showed in his model that there is a term which almost cancels the berry curvature term @xcite . ) on the other hand , kontani and yamada derived the intrinsic ahe due to the `` fermi surface term '' based on the linear - response theory for the first time @xcite .
they studied ahc in an orbitally degenerate periodic anderson model , and succeeded in explaining the ahe in hf systems ; @xmath26 below the coherent temperature @xmath27 , whereas @xmath28 above @xmath27 @xcite .
this results explain the experimental fact that the hall coefficients in hf systems are proportional to @xmath17 below @xmath27 .
miyazawa et al . proved that the intrinsic ahe also occurs in @xmath29-@xmath30 models with orbital degree of freedoms @xcite .
later , ahc s for fe and srruo@xmath31 were calculated based on the lda band calculations @xcite .
we still have to deepen the understanding of the mechanism of ahe to explain experimental results in transition ferromagnetic metals .
for example , a recent experiment by asamitsu et al . has revealed that the intrinsic ahc @xmath32 is observed in many ferromagnets for @xmath33 , whereas @xmath15 starts to decrease in proportion to @xmath34 and @xmath35 in bad metals where @xmath36 .
this drastic crossover is reminiscent of the ahe in heavy fermion systems discussed in ref .
therefore , a detailed study based on an appropriate model for transition metals is highly required .
in the present paper , we study the intrinsic ahc in a tight - binding model with @xmath0-orbitals ( @xmath37-orbitals ) , which originate from the @xmath38-orbitals @xmath39 in the tetragonal crystalline field .
the band structure of this model corresponds to @xmath40 and @xmath41 bands of sr@xmath42ruo@xmath43 @xcite .
we derive a general expression for the ahc valid for any damping rate @xmath4 , which enables us to study the ahc in bad metals .
we show that the ahc is mainly given by the fermi surface term for a wide range of @xmath4 since the kl s term is canceled by another fermi sea term in a metallic state .
we also find that the anomalous velocity due to @xmath44-@xmath45 hopping gives rise to a large ahc comparable to experimental values ( about @xmath46 ; @xmath47 being the lattice constant ) , which will be a main origin of a huge ahc in transition metal ferromagnets . in good metals ,
the intrinsic ahc is independent of @xmath48 as is well known .
however , it becomes proportional to @xmath3 in bad metals where @xmath4 is larger than the band - splitting around the fermi level , @xmath5 .
this crossover behavior of the ahc can explain a recent observation of the ahe by a. asamitsu et al . here , we shortly explain the mechanisms of both the intrinsic ahe and the extrinsic ahe in more detail : the intrinsic ahe ( both fermi surface and fermi sea terms ) originates from the interband transition of quasiparticles due to off - diagonal terms of the velocity @xmath49 and the orbital angular momentum @xmath50 .
( the diagrammatic expression for @xmath18 is given in fig .
[ fig : diagram ] . )
the off - diagonal velocity contains the `` anomalous velocity '' ( e.g. @xmath51 ) , which gives rise to the ahe .
since it is a purely quantum effect , it is difficult to find a simple classical analogue except for rashba - type 2d electron gas model @xcite .
this might be a reason why the intrinsic ahe had been discounted for a long time . in the present model ,
a large anomalous velocity naturally comes from the inter - orbital ( @xmath52-@xmath53 ) hopping . by this reason ,
atomic @xmath29-orbitals are necessary to reproduce a huge ahc in real ferromagnets .
on the other hand , the extrinsic ahe ( the skew - scattering mechanism and the side - jump mechanism ) happens even in a single - band model , although it is enhanced due to a multiband effect @xcite .
extrinsic ahcs due to skew - scattering mechanism and the side - jump mechanism are proportional to @xmath54 and @xmath55 , respectively .
the former is caused by spin - dependent asymmetric scattering around the impurity in the presence of spin - orbit coupling , and the latter is caused by lateral displacement of the wave - packet during the scattering @xcite .
diagrammatically , both skew - scattering and side - jump are expressed by _ current vertex corrections _
due to impurity potential , as shown in ref . @xcite .
in general , detailed knowledge on the impurity potential is needed for a quantitative study of the extrinsic ahe . in the present paper
, we do not study the extrinsic ahe since recent several experiments suggest that the skew scattering term is small even in good metals @xcite .
for a long time , it was believed that the ahe is independent of the _ self - energy correction _ due to impurity potential .
however , we show that the intrinsic ahc is reduced by the self - energy correction in bad metal . by this reason ,
change of the scaling law in the ahc at @xmath56 cm reported by asamitsu el at .
is well explained in terms of the intrinsic ahe .
in the present paper , we study a square lattice tight - binding model with @xmath52 and @xmath53 orbitals , which is a simplified version of the @xmath0-@xmath57 model .
miyazawa et al .
@xcite showed that the ahe exists in the latter model .
however , they could not derive an explicit expression for ahc . here , we derive explicit expressions for both the `` fermi surface term '' and the `` fermi sea term '' based on the present simplified model .
the @xmath58-coupling term is indispensable for the ahe in ferromagnets : in a ferromagnetic metal with @xmath59 , the @xmath58-coupling term @xmath60 is approximately proportional to @xmath61 , where @xmath62 is the coupling constant .
the ahe is caused by the inter - orbital transition of quasiparticles due to the off - diagonal elements of @xmath63 @xcite . in a paramagnetic metal under @xmath64 ,
on the other hand , zeeman term for the angular momentum , @xmath65 , gives rise to the ahe @xcite .
therefore , the ahe in paramagnetic metals and that in ferromagnetic ones are caused by the same transport mechanism , although the origins of magnetizations are different .
we consider that the present model with @xmath52- and @xmath53-orbitals describes a major part of the ahe in transition metal ferromagnets : in cubic or tetragonal crystals , @xmath66 is nonzero only when @xmath67 @xmath68 $ ] and @xmath69 @xmath70 $ ] . because levels of @xmath52 and @xmath53 orbitals are degenerate ,
the band structures composed of these orbitals are energetically close .
this fact will be favorable to the interband hopping of quasiparticles between @xmath52 and @xmath53 orbitals , which is indispensable for the ahe .
on the other hand , energy splitting between @xmath71-orbital and @xmath72-orbital is of the order of 1ev in square or tetragonal crystals .
therefore , the interband hopping of quasiparticles between @xmath71- and @xmath72-orbitals would be difficult . as a result
, the main contribution to the ahe in transition metal oxides will come from the inter - orbital transition between @xmath52- and @xmath53-orbitals .
this is the reason why we study the @xmath0-orbital tight - binding model . here ,
we represent the creation operator of an electron on @xmath73 ( @xmath74 ) orbital as @xmath75 ( @xmath76 ) .
the hamiltonian without @xmath58-coupling is given by @xmath77 , where @xmath78 and @xmath79 .
@xmath80 , @xmath81 and @xmath82 . here ,
@xmath83 and @xmath84 are the hopping integrals between nearest - neighbors and next - nearest - neighbors , respectively @xcite .
they are shown in fig .
[ fig : model ] . rotation .
this fact gives rise to the anomalous velocity . ] then , the velocity matrix is given by @xmath85 , where @xmath86 .
they are given by @xmath87 we should stress that the off - diagonal elements of @xmath88 , @xmath89 , is an odd - function of @xmath90 . in the same way , @xmath91 is an odd - function of @xmath92 .
they are called `` anomalous velocity '' . in later sections
, we will show that @xmath15 is proportional to @xmath93 , which can remain finite after the @xmath94-summation due to the anomalous velocity .
therefore , a sizable ahc is caused by @xmath95 and @xmath96 .
atomic @xmath29-orbitals give rise to the huge ahc in transition metal ferromagnets . to realize the ahe in ferromagnets , the atomic @xmath58-coupling is also necessary .
it is given by @xmath97 . because @xmath98 and @xmath99 , @xmath100 in the present basis is given by @xmath101 where @xmath62 is the coupling constant and @xmath102
is the pauli matrix for the orbital space .
note that @xmath103 in the present basis .
in fe , @xmath104mev @xcite .
hereafter , we put @xmath105 for the simplicity of calculation . and the velocity @xmath106 . ] the green function in the presence of atomic @xmath58-coupling is given by @xmath107 , which is given by @xmath108 where @xmath109 and @xmath110 , which is expressed as @xmath111 where @xmath112 represents the quasiparticle dispersion .
figure [ fig : fs ] shows the fermi surfaces for @xmath113 . in sr@xmath42ruo@xmath43 , @xmath114
the electron density per spin , @xmath115 , is set as @xmath116 and @xmath117 .
( note that @xmath118 in the present hubbard model according to slater - koster @xcite .
however , we assume @xmath119 because they are positive in sr@xmath42ruo@xmath43 due to the presence of @xmath57-orbital between the nearest @xmath120-sites @xcite . )
the splitting @xmath121 represents the minimum band - splitting ( @xmath122 ) measured from the the fermi surface of @xmath112-band . in fig .
[ fig : fs ] , @xmath123 represents the position of the minimum band - splitting , @xmath124 . and ( a ) @xmath125 .
here we put @xmath113 , which corresponds to sr@xmath42ruo@xmath43 .
, title="fig : " ] and ( a ) @xmath125 . here
we put @xmath113 , which corresponds to sr@xmath42ruo@xmath43 .
, title="fig : " ] here , we consider the quasiparticle damping rate @xmath126 , which is given by the imaginary part of the self - energy @xmath127 .
for simplicity , we assume that @xmath128 is diagonal with respect to orbital , and independent of momentum . then , retarded and advanced green functions with finite @xmath128 are given by @xmath129 where @xmath130 .
this assumption will be justified when the damping is caused by local impurities since the local green function @xmath131 is diagonal and independent of @xmath132 if @xmath133 .
[ offdiagonal term of @xmath134 is proportional to @xmath135 . ] in the born approximation , @xmath136 , where @xmath137 is the impurity concentration and @xmath138 is the impurity potential .
the born approximation is valid when @xmath138 is much smaller than the bandwidth @xmath139 . in appendix [ ap - offd ] ,
we derive the ahc in the born approximation by taking the offdiagonal term of @xmath140 into account correctly , and find that its effect is negligible . on the other hand , the assumption given in eqs .
( [ eqn : gr ] ) and ( [ eqn : ga ] ) may not be valid in strongly correlated systems . in appendix
[ ap - mom ] and [ ap - band ] , we study the intrinsic ahc when @xmath128 depends on momentum and band index , and find that the ahc could be changed prominently by these influences .
in this section , we derive the intrinsic ahc @xmath18 ( hereafter , we drop the superscript @xmath47 ) and the longitudinal conductivity @xmath25 for finite quasiparticle damping rate @xmath128 based on the linear - response theory @xcite . here
, we take the self - energy correction due to correlation , which causes finite quasiparticle damping rate . on the other hand
, we drop all the current vertex correction ( cvc ) .
the cvc due to impurities with @xmath58-coupling give rise to the skew scattering @xcite .
however , cvc due to local impurities vanishes identically .
note that the cvc due to the coulomb interaction does not cause the skew scattering @xcite .
hereafter , we put the renormalization factor @xmath141 because @xmath142 exactly cancels in the final formula of the ahc . from now on
, we assume a complete ferromagnetic state with @xmath143-spin electrons only , i.e. , @xmath144 .
hereafter , we drop the factor @xmath145 ( @xmath146 being the plank constant ) in @xmath147 to simplify expressions . according to streda @xcite ,
@xmath147 is given by @xmath148 , \nonumber \\
\label{eqn : sigma - i } \\ \s_{\mu\nu}^{ii}&= & -\frac{1}{2}\sum_{\k,\a\a'\b\b ' } \int\frac{d\e}{2\pi } v_\mu^{\a'\a } v_\nu^{\b'\b } f(\w ) \nonumber \\ & & \times \left [ \frac{\d}{\d\e}g_{\a\b'}^r \cdot g_{\b\a'}^r -g_{\a\b'}^r \frac{\d}{\d\e } g_{\b\a'}^r \right .
\nonumber \\ & & \ \ \ \ \ \ \left .
-\frac{\d}{\d\e}g_{\a\b'}^a \cdot g_{\b\a'}^a + g_{\a\b'}^a \frac{\d}{\d\e } g_{\b\a'}^a \right ] , \label{eqn : sigma - ii}\end{aligned}\ ] ] where @xmath149 and @xmath150 are given by eqs .
( [ eqn : vx ] ) , ( [ eqn : vy ] ) , ( [ eqn : gr ] ) and ( [ eqn : ga ] ) . @xmath151 is the fermi distribution function .
the diagrammatic expression is given in fig .
[ fig : diagram ] ( a ) .
we call @xmath152 and @xmath153 the `` fermi surface term '' and the `` fermi sea term '' , respectively , according to literature .
the original kl s paper @xcite studied only a part of the fermi sea term ( @xmath154 given in eq .
( [ eqn : xy - ii2 ] ) ) , which is equivalent to the berry curvature term .
on the other hand , kontani and yamada @xcite derived the fermi surface term @xmath155 in the @xmath156 periodic anderson model .
they also showed that the fermi sea term is absent in this model .
in general models , however , both contributions exist as recognized by luttinger @xcite .
recently , a detailed analysis for both terms was given in ref .
@xcite for graphene .
however , comparison of each term was not complete for @xmath29-electron systems . in the present paper
, we show that @xmath155 is much larger than @xmath157 in metallic systems for a wide range of @xmath126 .
we note that the normal hall conductivity due to the lorentz force is proportional to @xmath158 . in this case
, @xmath155 is much larger than @xmath159 in good metals where @xmath160 is very large . on the other hand ,
the intrinsic ahc is of the order of @xmath161 . by this reason
, both @xmath155 and @xmath157 could be the same order . .
( b ) first order term with respect to the atomic @xmath58-coupling in terms of the band - diagonal representation . ] from now on , we calculate the intrinsic ahc for @xmath162 .
first , we take the summation over @xmath163 in eqs . (
[ eqn : sigma - i ] ) and ( [ eqn : sigma - ii ] ) . after taking the @xmath94-summation ,
only the terms @xmath164 remain finite . using the square lattice symmetry of the present model
, we obtain that @xmath165 , \label{eqn : sxy - ii}\end{aligned}\ ] ] where @xmath166 . here , it is essential that @xmath167 is the totally symmetric representation ( @xmath168 ) , and it does not vanish after the @xmath94-summation . note that the term proportional to @xmath169 in @xmath170 vanishes identically . here
, we derive the expressions for @xmath170 and @xmath157 at @xmath171 .
then , the @xmath172-integration in eq .
( [ eqn : sxy - ii ] ) can be performed using the relation @xmath173 the final result for the intrinsic ahc is given by @xmath174 below , we perform the numerical calculation for the ahc using eqs .
( [ eqn : xy - i ] ) , ( [ eqn : xy - ii1 ] ) and ( [ eqn : xy - ii2 ] ) . in the same way
, we calculate the longitudinal conductivity for @xmath175 . in this case ,
( [ eqn : sigma - ii ] ) vanishes identically , and only the fermi surface term ( [ eqn : sigma - i ] ) remains finite . after taking the @xmath94-summation and using the square lattice symmetry of the present model
, we obtain the expression for @xmath25 at @xmath171 as follows : @xmath176 \right .
+ 2(v_x^{xy})^2 \nonumber \\ & & \ \ \times [ ( \xi^{xy})^2 - \l^2 + ( \mu-\xi_y)(\mu-\xi_x)+\gamma^2 ] \nonumber \\ & & \left .
+ 4v_x^{xx}v_x^{xy } \xi^{xy}(\mu-\xi_y ) \right\ } /d^r(0)d^a(0 ) , \label{eqn : xx - i1 } \\
\s_{xx}^{ib}&= & \frac{-1}{2\pi } { \rm re } \sum_\k \left\ { ( v_x^{xx})^2(\mu-\xi_y + i\gamma)^2 \right
. + 2(v_x^{xy})^2 \nonumber \\ & & \ \times \left [ ( \xi^{xy})^2 - \l^2 + ( \mu-\xi_y+i\gamma)(\mu-\xi_x+i\gamma ) \right ] \nonumber \\ & & \left . + 4v_x^{xx}v_x^{xy } \xi^{xy}(\mu-\xi_y+i\gamma ) \right\ } /(d^r(0))^2 .
\label{eqn : xx - i2 } \end{aligned}\ ] ] note that @xmath177 vanishes identically . in the case of @xmath178
, @xmath179 diverges as @xmath180 whereas @xmath181 is finite .
therefore , @xmath182 in good metals .
however , @xmath179 and @xmath181 become the same order when @xmath128 is very large .
we perform the numerical calculation for the @xmath25 using eqs .
( [ eqn : xx - i1 ] ) ( [ eqn : xx - i2 ] ) in later sections . before performing the numerical study in [ numerical ]
, we analyze eqs .
( [ eqn : xy - i ] ) , ( [ eqn : xy - ii1 ] ) and ( [ eqn : xy - ii2 ] ) when @xmath128 is very small in detail : in this case , @xmath183 substituting above equations into ( [ eqn : xy - i])-([eqn : xy - ii2 ] ) , we obtain the following relation for @xmath178 : @xmath184 according to eq .
( [ eqn : ap - xy - ii2 ] ) , the main contribution to @xmath154 comes from an area near @xmath123 in fig [ fig : fs ] , if @xmath185 and @xmath186 are satisfied .
when @xmath187 is small , then @xmath188 .
in this case , @xmath189 note that this relation will not be satisfied in a special case where the fermi level lies between a narrow bandgap . in the next section
, we see that eq .
( [ eqn : approx ] ) holds very well in the present model . here , we summarize the obtained results in this section . +
( i ) @xmath154 was first recognized by karplus and luttinger @xcite .
it can be rewritten in terms of the summation of the berry curvature @xcite .
+ ( ii)@xmath190 is another `` fermi sea term '' , although its final expression ( [ eqn : ap - xy - ii1 ] ) is written like a fermi surface term .
we find that @xmath157 is very small in usual metals because @xmath190 and @xmath154 almost cancels each other .
+ ( iii ) the fermi surface term is canceled by @xmath190 in the clean limit , which is also recognized in 2d dirac model @xcite .
however , this cancellation is imperfect when @xmath128 is finite , as shown in the next section .
+ several previous works assumed that the intrinsic ahc is given by @xmath154 @xcite .
one might consider that this assumption is justified by the present calculation for @xmath178 .
however , this assumption is not always guaranteed as we will see in the next section : when @xmath128 is as large as @xmath5 , then ( a ) @xmath191 still holds whereas ( b ) @xmath24 becomes quite different from @xmath154 . in appendix
[ ap - mom ] and [ ap - band ] , we study the intrinsic ahc in the general case where @xmath128 is @xmath94-dependent or it depends on band index . in these cases ,
statement ( a ) and ( b ) are also true .
therefore , overall behavior of @xmath24 for a wide range of @xmath128 is well expressed by the fermi surface term @xmath155 , not by @xmath154 .
this is an important result of this paper , which is obtained by considering all the terms contributing to the intrinsic ahc . finally , we comment that both @xmath192 and @xmath193 change their signs under the gauge transformation @xmath194 .
the ahc is invariant under this gauge transformation .
in this section , we perform the numerical calculation for both @xmath24 and @xmath25 at @xmath171 , assuming a complete ferromagnetic state where @xmath195 and @xmath196 . in this case ,
( we put @xmath105 hereafter . )
the main purpose of this section is to elucidate both the filling ( @xmath115 ) and damping rate ( @xmath128 ) dependences of the ahc .
we perform the @xmath94-summations in eq .
( [ eqn : xy ] ) for @xmath24 and eq .
( [ eqn : xx ] ) for @xmath25 numerically , dividing the brillouin zone into 5000@xmath1985000 meshes .
the unit of conductivity in this section is @xmath46 , where @xmath146 is the plank constant and @xmath47 is the unit cell length ( inter - layer distance in 2d systems ) . if we assume the length of unit cell @xmath47 is 4 , @xmath199 . ) and @xmath170 as a function of @xmath200 .
it is shown that @xmath201 is well satisfied .
@xmath199 if we put the unit cell length @xmath202 . @xmath24 is not a monotonic function of @xmath203 , and changes its sign at @xmath204 ] figure [ fig : ahe - n ] shows the @xmath115-dependence of @xmath24 for @xmath205 .
@xmath206 corresponds to a ferromagnetic band insulators . in the present model , @xmath207
here , we put @xmath208 , which is sufficiently smaller than the minimum band - splitting measured from the fermi surface , @xmath5 , which is shown in fig .
[ fig : fs ] .
the bandwidth of the present model is approximately @xmath209 .
we put the @xmath58-coupling constant @xmath135 as @xmath210 .
if we assume @xmath211k , it corresponds to @xmath212k , which are realistic values in transition metals . in the present study
, @xmath213 exceeds @xmath214 at @xmath215 .
the obtained magnitude of @xmath213 is comparable with experimental value in fe @xcite and in srruo@xmath31 @xcite .
figure [ fig : lambda ] shows the @xmath62-dependence of @xmath213 for @xmath216 and 1.4 .
@xmath213 is approximately proportional to @xmath62 below @xmath217 , whereas it tends to saturate when @xmath62 is as large as @xmath218 since @xmath219 increases with @xmath62 .
@xcite . in fig .
[ fig : ahe - n ] , @xmath24 is positive ( negative ) for @xmath220 ( @xmath221 ) ; @xmath222 at @xmath223 .
the reason for the sign change of @xmath222 is the following : according to eq .
( [ eqn : xy ] ) , the main contribution for @xmath224 comes from an area near @xmath225 .
therefore , @xmath226 .
because @xmath227 , the sign of @xmath24 changes to be negative when @xmath228 exceeds @xmath229 . the sign change of the ahc as a function of @xmath203 is actually observed in srruo@xmath31 .
-dependence of @xmath24 for @xmath125 and @xmath116 . ] next , we examine the @xmath128-dependence of the ahc .
figure [ fig : logsxy-123 ] shows the total ahc ( @xmath24 ) , @xmath170 , @xmath230 and @xmath154 for @xmath125 .
we see that all of them are @xmath128-independent when @xmath231 . on the other hand
, they start to decrease with @xmath128 when @xmath232 : this is easily recognized from the functional forms of eqs .
( [ eqn : xy - i])-([eqn : xy - ii2 ] ) .
we see that @xmath170 is almost equal to @xmath230 and @xmath233 when @xmath231 , as discussed in the previous section . on the other hand , @xmath201 whereas @xmath24 is quite different from @xmath154 for @xmath234 . as a result
, the fermi surface term @xmath155 succeeds in reproducing the overall behavior of the ahc for a wide range of @xmath128 .
@xmath24 is proportional to @xmath235 for @xmath236 , which can be recognized by eq .
( [ eqn : xy - i ] ) .
this crossover behavior of @xmath24 at @xmath237 was first pointed out by ref .
@xcite . they found that @xmath24 is proportional to @xmath158 for @xmath238 in @xmath156 periodic anderson model ( pam ) .
( @xmath239 is the energy of @xmath240-electrons . )
the different @xmath128-dependence of @xmath24 between in @xmath29-@xmath30 model and in pam comes from the fact that the conduction bandwidth @xmath241 is much wider than the heavy quasiparticle band in pam . in the pam ,
the quasiparticle damping is given by @xmath242 , where @xmath22 is the @xmath243-@xmath240 mixing term and @xmath244 is the imaginary part of the @xmath240-electron self - energy .
because @xmath245 saturates at @xmath246 and the relation @xmath247 holds in the pam , @xmath245 will always be smaller than @xmath241 even if @xmath248 .
-dependences of @xmath249 , @xmath250 , @xmath251 and @xmath252 , respectively .
@xmath24 shows a crossover behavior at @xmath253 .
it is shown that @xmath170 reproduces a overall behavior of the total @xmath24 for a wide range of @xmath128 . ]
we also study the @xmath128-dependences of the longitudinal conductivity @xmath25 .
figure [ fig : logsxx-12 ] shows that the relations @xmath254 and @xmath255 hold well for a wide range of @xmath128 . the conductivity @xmath25 does not show any clear crossover behavior at @xmath237 . when @xmath236 , @xmath25 seems proportional to @xmath256 , and @xmath181 is as large as @xmath179 .
although eqs .
( [ eqn : xx - i1 ] ) and ( [ eqn : xx - i2 ] ) suggest that @xmath257 when @xmath258 , we could not find such a behavior for @xmath259 .
-dependences of @xmath179 , @xmath181 and @xmath260 . ]
finally , we study the relation between the ahc and the resistivity @xmath261 .
figure [ fig : logsxy ] shows that @xmath24 is independent of @xmath2 when @xmath262 $ ] . on the other hand
, @xmath24 starts to decrease for @xmath263 in proportion to @xmath3 .
this crossover behavior of @xmath24 at @xmath264 is observed universally by recent experiments on various transition metal ferromagnets by asamitsu et al.@xcite . and
@xmath261 for @xmath265 and @xmath266 .
@xmath267 corresponds to @xmath268 .
it reproduces well the experimental universal `` asamitsu plot '' for various transition metal ferromagnets , where @xmath269 for @xmath270 , whereas @xmath271 for @xmath272 . ]
in this section , we analyze the intrinsic ahc in terms of the band - diagonal representation , which has been analyzed by many authors . we note that this representation is useful for the analysis of the ahc only when @xmath273 is orbital independent . in this representation , according to eqs .
( [ eqn : sigma - i ] ) and ( [ eqn : sigma - ii ] ) , @xmath274 at zero temperature . here ,
@xmath275 and @xmath276 are the band suffices , and we dropped the diagonal terms @xmath277 because they vanish identically . after performing the @xmath278-integral in @xmath157 ,
the fermi sea term is expressed as @xmath279 : @xmath280 when @xmath128 is very small , they are given by @xmath281 because @xmath282 and @xmath283 , @xmath154 is equivalent to eq .
( 2.16 ) of ref .
@xcite , or eq . ( 3 ) of ref .
@xcite . based on eq .
( [ eqn : app - ii2 ] ) , when the fermi energy lies in a gap , @xmath154 is quantized when the fermi energy lies in a gap @xcite . in the metallic state , on the other hand , @xmath154 is rewritten as the fermi surface term due to the partial integral .
off course , we can reproduce eqs .
( [ eqn : ap - xy - i])-([eqn : ap - xy - ii2 ] ) by applying eqs .
( [ eqn : app - i])-([eqn : app - ii2 ] ) to the present tight - binding model given in [ model ] . in the band - diagonal representation ( @xmath284 ) , we can show that the off - diagonal velocity is given by @xmath285 and @xmath286 .
therefore , @xmath287 where @xmath288 and @xmath289 . substituting this result into eqs .
( [ eqn : app - i])-([eqn : app - ii2 ] ) , we obtain the same results given in [ ahe ] .
in the present paper , we studied the mechanism of the intrinsic ahe in transition metal ferromagnets based on the @xmath0-orbital tight - binding model .
the origin of the anomalous velocity is the hopping integral between @xmath52 and @xmath53 orbitals , @xmath192 . by virtue of it
, we could reproduce a typical experimental value of the ahc in ferromagnets ; @xmath290 .
thus , the anomalous velocity due to atomic @xmath29-orbitals will be the main origin of the ahe in transition metal oxides .
this fact has been overlooked in previous theoretical works based on electron gas models without atomic orbitals .
we note that the present model breaks the parity symmetry [ @xmath291 due to @xmath292 .
this is a necessary condition for a spontaneous hall effect @xcite . in accord with the present study
, paramagnetic compound ca@xmath293sr@xmath294ruo@xmath43 shows large ahe under the magnetic field @xcite .
its magnitude is comparable with the large ahe in @xmath240-electron systems ( such as upt@xmath31 ) which originates from the anomalous velocity due to atomic @xmath240-orbitals@xcite .
the fermi surfaces of this compound are composed of @xmath38-orbital ( @xmath52 , @xmath53 , @xmath72-orbitals ) ; @xmath128-band composed of @xmath72-orbital is absent in the present ( @xmath52 , @xmath53)-tight - binding model .
we will study the ahe of this compound in more detail based on the @xmath38-orbital tight - binding model , which reproduces the bandstructure of ( ca , sr)@xmath42ruo@xmath43 accurately @xcite .
we derive a general expression for the ahc valid for any quasiparticle damping rate @xmath245 , by performing the analytic continuation carefully .
[ equations ( [ eqn : xy - i ] ) , ( [ eqn : xy - ii1 ] ) and ( [ eqn : xy - ii2 ] ) for @xmath0-orbital model , and eqs .
( [ eqn : ap - i ] ) , ( [ eqn : ap - ii1 ] ) and ( [ eqn : ap - ii2 ] ) for a general model . ] using the general expression , we succeeded in explaining the experimental crossover behavior of the ahc in bad metals ; @xmath24 is constant for @xmath295 , whereas @xmath296 for higher resistivity .
this overall behavior is mainly given by the fermi surface term , @xmath170 , whose importance was intensively studied in ref .
@xcite .
we stress that the intrinsic ahc in the present model is not a monotonic function with respect to @xmath297 . in partial ferromagnets ,
the total ahc is given by @xmath298 where @xmath299 is given by eq .
( [ eqn : xy ] ) , and @xmath300 is the chemical potential for the electron density @xmath301 .
we found that the sign - change of @xmath24 occurs in correspondence with sign of @xmath95 at @xmath123 .
we expect that this result will explain the sign - change of the ahe in srruo@xmath31 as a function of the magnetization .
we comment that authors of ref .
@xcite found that a large value of ahc appears from @xmath154 when fermi level lies inside a narrow " anticrossing band gap , and they suggested that this mechanism accounts for a huge ahc in ferromagnets .
however , this condition will be satisfied only for a narrow range of magnetization
. the present model with @xmath0-orbitals can give an enough magnitude of the ahc , although band crossings are absent ( except at @xmath302 and @xmath303 ) .
we stress that the ahc in the present model shows a sign change with respect to @xmath203 , even if the topology of the fermi surface is unchanged ; see fig .
[ fig : ahe - n ]
. the relation @xmath304 ( berry curvature term ) will not hold in usual metallic compounds : for example , in bad metals where @xmath128 is as large as the minimum band - splitting measured from the fermi surface , @xmath5 , the relations @xmath305 ( fermi surface term ) and @xmath306 hold well for a wide range of @xmath128 as shown in fig .
[ fig : logsxy-123 ] . even in good metals ,
the relation @xmath304 is also invalid when ( a ) @xmath128 is @xmath94-dependent , and/or ( b ) @xmath128 depends on band index , i.e. , @xmath307 : in these cases , the fermi surface term deviates from eq .
( [ eqn : ap - xy - i ] ) as shown in appendix [ ap - mom ] and [ ap - band ] . in summary
, the relation @xmath304 will hold only in good metals as well as when @xmath308 is independent of @xmath275 , @xmath94 and @xmath172 on the other hand , the relation @xmath309 will be universal in real metallic systems , since the fermi sea terms almost cancel each other except for a special situation where @xmath310 lies inside a narrow anticrossing band gap .
this result is consistent with recent theoretical work on metallic graphene @xcite . for a quantitative study of the intrinsic ahc
, however , we have to calculate @xmath155 , @xmath190 and @xmath154 on the same footing .
this is also important conclusion in this paper .
finally , we comment that the present @xmath0-tight binding model in a paramagnetic state shows a finite spin hall conductivity ( shc ) @xmath311 , that is , @xmath312-spin current along @xmath313-axis occurs under the electric field along @xmath314-axis .
@xmath311 is given by @xmath315 times eqs .
( [ eqn : xy ] ) . as shown in fig .
[ fig : ahe - n ] , @xmath311 reaches @xmath316 \sim 600 \ [ \hbar e^{-1 } \cdot \omega^{-1}{\rm cm}^{-1}]$ ] for @xmath3174 , which is almost one order of magnitude larger than typical shc in semimetals @xcite . in later publications
, we will study the spin hall effect in more detail @xcite asamitsu et al . examined the experimental relation between the ahc and the resistivity @xmath2 in various ferromagnets .
they observed the dissipation - less intrinsic ahc in the intermediate conducting region with @xmath270 . in the bad metal region , @xmath24 is proportional to @xmath3 , which is well explained in the present study .
we found this crossover at @xmath318 occurs when @xmath319 . according to fig .
[ fig : logsxx-12 ] , @xmath268 corresponds to @xmath320k .
another crossover to the good metal region ( @xmath321 ) seems to be rather complex : it might be due to the extrinsic ahe or a phenomenon related to the normal hall effect .
recently , s. onoda et al .
@xcite calculated both the intrinsic ahc @xcite and the extrinsic ahc @xcite in a rashba - type 2d electron gas ( 2deg ) model .
they claimed that the extrinsic - intrinsic crossover occurs when @xmath319 , and tried to explain the crossover at @xmath20 observed by asamitsu .
however , it will not be true because @xmath322 corresponds to @xmath323k , which is too small for a realistic value of @xmath5 .
( see fig .
[ fig : logsxx-12 ] . )
we stress that inoue et al .
@xcite proved that the intrinsic ahe in a rashba model almost vanishes due to the cancellation by the current vertex correction ( cvc ) by impurities . on the other hand
, we have checked that the cvc for anomalous velocity is absent in the present @xmath0-orbital model within the born approximation , when the impurity potential is delta - functional . by this reason ,
the intrinsic ahc obtained in the present work is not modified by the cvc .
we conclude that the present model with atomic @xmath29-orbital degrees of freedom is essential for a realistic study of ahe in ferromagnets . in paramagnetic heavy fermion ( hf )
systems , the hall coefficients takes huge values due to the ahe , since the uniform susceptibility @xmath12 in hf is prominently enhanced by the strong coulomb interaction . in hf systems ,
the large quasiparticle damping rate @xmath324 comes from the strong electron correlation ( or kondo resonance ) . in the early stage ,
coleman et al @xcite and fert and levy @xcite developed theories of the extrinsic ahe : developed theories of extrinsic ahe : they studied the extrinsic mechanism based on the @xmath240-electron impurity anderson models with @xmath29-orbital channel , and predicted the relation @xmath325 above @xmath27 .
they assumed that the impurity anderson models is applicable for the study of ahe in periodic systems if @xmath326 .
on the other hand , kontani and yamada found that the @xmath156 periodic anderson model ( pam ) gives a large intrinsic ahc @xcite .
they predict that @xmath26 below @xmath27 , whereas @xmath28 above @xmath27 .
we expect this intrinsic ahe is widely observed since the @xmath156 pam is a well - established effective model for ce - based hf systems : for example , it could explain a large van - vleck susceptibility in kondo insulator @xcite .
there remain a lot of future works to be done about the ahe in hf systems .
for example , nakajima et al . found that the ahc in two - dimensional hf system , cemin@xmath327 ( m = co , rh , ir ) is negligibly small @xcite . in this compound ,
strong temperature dependence of @xmath328 is given by the normal hall effect , due to the effect of the current vertex corrections ( or backflow ) in nearly antiferromagnetic fermi liquids @xcite .
in addition , an interesting field - direction dependence of ahc is found in cecu@xmath329 or cecu@xmath330au@xmath331 @xcite .
theoretical studies on these experimental results will give us important hints to understand the electronic states .
we are grateful to j. inoue , y. tanaka , d.s .
hirashima , m. sato and h. sato for enlightening discussions .
we also thank h. fukuyama for sending us his thesis @xcite .
in previous sections , we studied the ahc by assuming that im@xmath332 is diagonal with respect to atomic orbital .
this assumption seems to be justified since @xmath333 when @xmath58-coupling constant @xmath62 is much smaller than bandwidth @xmath139 .
[ when @xmath334 , @xmath335 since the local green function @xmath336 is diagonal due to the orthonormality of @xmath29-orbitals .
] here , we derive the ahc within the born approximation , by taking both the diagonal and off - diagonal elements of im@xmath332 . as a result , it is confirm that the off - diagonal element is negligible in calculating the ahc in the present model . in the born approximation ,
the green function is given by @xmath337 where @xmath128 , @xmath338 and @xmath146 come from the self - energy correction in the self - inconsistent born approximation : @xmath339 where @xmath137 is the impurity concentration and
@xmath138 is the impurity potential .
therefore , @xmath340 , where @xmath341 is the dos at the fermi level . in the same way , we can estimate that @xmath342 .
we can derive the expression for the ahc by inserting eqs .
( [ eqn : gr2 ] ) and ( [ eqn : ga2 ] ) into the streda formula in [ ahe ] , in the same way of deriving eqs .
( [ eqn : sxy - i ] ) and ( [ eqn : sxy - ii ] ) . using the square lattice symmetry of the present model , the fermi surface term at zero temperature is given by @xmath343 , \nonumber \\
\label{eqn : sxy - born}\end{aligned}\ ] ] since the main contribution in the @xmath94-summation comes from @xmath344 , where @xmath123 represents the position of the minimum band - splitting @xmath5 , as shown in fig .
[ fig : fs ] . since @xmath345 at @xmath346 , the second term of eq .
( [ eqn : sxy - born ] ) is negligible .
therefore , the correction due to off - diagonal element of the self - energy correction , @xmath347 , is very small . as a result ,
( [ eqn : sxy - i ] ) is justified in the born approximation . ) .
self - energy correction due to impurity scattering is not diagonal when @xmath348 .
( b ) and ( c ) represent the born and the second born approximation , respectively .
the latter gives the skew scattering extrinsic ahe when the impurity potential is non - local .
( d ) represents the side jump extrinsic ahc . ]
figure [ fig : diagram2 ] ( a ) represents the intrinsic ahc given in eq .
( [ eqn : sxy - born ] ) .
in good metals , the diagonal self - energy correction does not affect the ahc , and the ahc due to off - diagonal self - energy correction is negligible as we discussed above .
in addition , current vertex correction ( cvc ) term ( b ) also exists in the born approximation .
figure ( c ) and ( d ) represent other cvc terms which gives the extrinsic ahc : ( c ) gives the skew scattering extrinsic ahc that is proportional to @xmath25 , and ( d ) represents the side - jump extrinsic ahc term that is proportional to @xmath349 , respectively .
the latter is caused by the lateral displacement of the wave - packet during the scattering , which is theoretically expressed by the modification of velocity operator due to impurity potential @xcite . in the present model , ( b)-(d )
vanish identically in the case of local impurity potential . in contrast , extrinsic ahc occurs by local impurity potential in 2d electron gas models when the @xmath58-coupling term is @xmath94-dependent @xcite .
in [ ahe ] , [ numerical ] and [ banddiagonal ] , we have neglected the momentum dependences of the damping rate @xmath128 to simplify the discussion .
however , this assumption is not realistic because im@xmath350 is usually @xmath94-dependent . in this appendix
, we study the ahe by taking the @xmath94-dependence of @xmath128 into account
. then , we have to distinguish @xmath351 and @xmath352 .
we show that the expression for @xmath170 in eq .
( [ eqn : ap - xy - i ] ) could be changed prominently when the @xmath94-dependence of @xmath245 is taken into account . along with the derivation of eq .
( [ eqn : sxy - i ] ) , we obtain the expression for @xmath170 in the present case : @xmath353 where @xmath354 when both @xmath355 and @xmath356 are very small , @xmath357 in this case , eq .
( [ eqn : b - i ] ) is given by @xmath358 \label{eqn : b - ig}\end{aligned}\ ] ] when @xmath359 is momentum dependent , @xmath360 except for @xmath361 . therefore , eq . ( [ eqn : b - ig ] ) is not equal to eq .
( [ eqn : ap - xy - i ] ) even in good metals ( @xmath362 ) . in general , the momentum dependence ( and the band - dependence ) of @xmath128 could give rise to a nontrivial modulation of the ahc due to the fermi surface term , @xmath170 .
this effect will be prominent in strongly correlated electron systems like high-@xmath363 cuprates and heavy fermion systems . on the other hand , the fermi sea term @xmath157 , which is given by eqs .
( [ eqn : ap - xy - ii1 ] ) and ( [ eqn : ap - xy - ii2 ] ) , is satisfied when @xmath362 , even if @xmath364 .
as shown in appendix [ ap - mom ] , we have shown that the expression for fermi surface term @xmath170 given in eq .
( [ eqn : ap - xy - i ] ) is modified by the @xmath94-dependence of @xmath245 . in this section , we explain that eq .
( [ eqn : ap - xy - i ] ) is also changed when @xmath245 depends on bands .
this fact was already pointed out in appendix d of ref .
@xcite . here
, we derive the first order term of @xmath170 with respect to the @xmath58-coupling , and correct a mistake in appendix d of ref . @xcite . by extracting a single @xmath58-coupling term from eq .
( [ eqn : ap - i ] ) , we obtain @xmath365 , \end{aligned}\ ] ] where @xmath366 and @xmath367 are the green function and the velocity without @xmath58-coupling term . the diagrammatic expression is given in fig .
[ fig : diagram ] ( b ) . using the relation @xmath368 and @xmath369
, we obtain that @xmath370 .
\label{eqn : apc - i}\end{aligned}\ ] ] when @xmath371 is very small , @xmath372 .\end{aligned}\ ] ] in this case , eq .
( [ eqn : apc - i ] ) is given by @xmath373 \nonumber \\ & & \times \frac{i\langle l|l_z|m\rangle}{(e_\k^l - e_\k^m)^2 } \left [ v_x^{l l}v_y^{l m } - v_x^{l m}v_y^{l l } \right ] .
\label{eqn : apc - i2}\end{aligned}\ ] ] in summary , @xmath170 depends on the ratio of @xmath374 .
@xmath170 could take a huge value if @xmath375 differs much from band to band .
this fact was already pointed out by ref
. @xcite . on the other hand , eqs .
( [ eqn : ap - xy - ii1 ] ) and ( [ eqn : ap - xy - ii2 ] ) for the fermi sea term @xmath157 is satisfied when @xmath376 , even if @xmath377 . the present result will be valid even if the @xmath58-coupling is treated unperturbatively , as done in [ model ] - [ numerical ] .
in fact , fig . [
fig : lambda ] shows that @xmath24 is approximately linear - in-@xmath135 for @xmath378 . in appendix d of ref .
@xcite , @xmath379 in @xmath380 was replaced with @xmath381 by mistake .
thus , eq . ( [ eqn : apc - i2 ] ) is not equal to the kl s term which is given by [ eq .
( 3.5)]@xmath382 in ref .
@xcite with @xmath383 .
this difference is natural since the former and the latter are the fermi surface term and the fermi sea term , respectively . | to elucidate the origin of anomalous hall effect ( ahe ) in ferromagnetic transition metals , we study the intrinsic ahe based on a multi - orbital ( @xmath0 ) tight - binding model . we find that a large anomalous velocity comes from the off - diagonal ( inter - orbital ) hopping . by this reason
, the present model shows a large intrinsic anomalous hall conductivity ( ahc ) which is compatible with typical experimental values in ferromagnets [ @xmath1 , without necessity to assume a special band structure at the fermi level . in good metals where @xmath2 is small ,
the intrinsic ahc is constant ( dissipation - less ) as found by karplus and luttinger . in bad metals , however , we find that the ahc is proportional to @xmath3 when @xmath4 is larger than the minimum band - splitting measured from the fermi level , @xmath5 .
this crossover behavior of the intrinsic ahe , which was first derived in j. phys .
. jpn .
* 63 * ( 1994 ) 2627 , is recently observed in various ferromagnetic metals universally by a. asamitsu et al .
we also stress that the present @xmath0-tight binding model shows a huge spin hall effect in a paramagnetic state . |
the x - ray , gev gamma - ray , and tev gamma - ray skies have been deeply investigated by various telescopes . however , the mev gamma - ray sky has not been extensively studied yet after the imaging compton telescope ( comptel ) onboard the _ compton gamma - ray observatory _ ( _ cgro _ ) satellite scanned the sky in 1990s @xcite .
this means that huge discovery space is buried in this energy range .
measurement of mev gamma rays is hampered by huge background events . under this difficulty ,
the comptel achieved the sensitivity of @xmath4 in the mev band and discovered @xmath563 gamma - ray sources ( 32 of them were gamma - ray bursts ) at 0.7530 mev @xcite .
the comptel has found 10 extragalactic active galactic nuclei ( agns ) and 5 unidentified high latitude objects @xcite . to open a new observational window in the mev sky ,
various missions have been proposed .
astro - h _
@xcite , which will be launched in 2015 , will have a sub - mev instrument , the soft gamma - ray detector ( sgd , * ? ? ?
* ; * ? ? ?
* ) covering at 40600 kev
. the sgd will provide precise spectra with higher sensitivity than previous instruments by introducing the concept of narrow - field - of - view compton telescope which can dramatically reduce background events @xcite .
other instruments are also proposed such as cast @xcite , dual @xcite , grips @xcite , smile @xcite , cosi @xcite , and astrogam . moreover
, the instrumental performance were tested with ballon experiments @xcite .
these future missions will unveil the mev sky in near future .
in addition , a new image reconstruction algorithm that has improved sensitivity to multiple point - like sources has been proposed toward those future all - sky missions @xcite .
the new era of the mev gamma - ray astronomy is approaching .
a fundamental question to those missions is how many sources one can expect with their designed instruments . in the extragalactic sky ,
the dominant population in the mev sky is still uncertain because the origin of the cosmic mev gamma - ray background from 0.3 mev to 100 mev is not well established ( see * ? ? ? * for a recent review ) .
seyferts which make up the cosmic x - ray background ( cxb ) can not explain this mev background because of the assumed cutoff at a few hundred kev ( e.g. * ? ? ?
* ; * ? ? ?
above 100 mev , the total gamma - ray background is known to be predominantly composed of blazars ( e.g. * ? ? ? * ; * ? ? ? * ) , star - forming galaxies ( e.g. * ? ? ? * ) , and radio galaxies ( e.g. * ? ? ? * ) .
although the mev background may comprises these gev gamma - ray populations as well , the lower energy part of the mev background spectrum is smoothly connected to the cxb spectrum and shows softer ( photon index of @xmath6 2.8 * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) than the gev component ( photon index of @xmath6 2.4 * ? ? ?
* ) , indicating their different origins .
several candidates have been considered as the origin of the mev background .
one was the nuclear - decay gamma rays from type ia supernovae @xcite .
however , the measurements of the cosmic type ia supernovae rates show that the rates are not enough to explain the observed flux ( e.g. * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
seyferts may naturally explain the mev background up to a few tens of mev and the smooth connection to the cxb ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* hereinafter i08 ) .
non - thermal electrons in coronae can generate the power - law spectrum in the mev band after the thermal cut - off via the compton scattering of disk photons ( i08 ) .
flat - spectrum radio quasars ( fsrqs ) whose peak in the spectrum locates at mev energies @xcite are also expected to significantly dominate the mev background ( * ? ? ?
* hereinafter a09 ) .
radio galaxies have been also discussed as one of the origins of the mev background @xcite .
recent studies have revealed that both of lobe @xcite and core @xcite emissions from radio galaxies could contribute only @xmath7% of the mev gamma - ray background flux .
dark matters has also been discussed as the origin of the mev background ( e.g. * ? ? ?
* ; * ? ? ?
those mev mass dark matter particle candidates are less natural than gev - tev dark matter candidates .
therefore , seyferts and fsrqs can be regarded as potential astrophysical origins of the mev background .
the purpose of this paper is to investigate prospects of extragalactic observations by future mev instruments , especially of statistical aspects of agns .
we focus only on seyferts and fsrqs in this paper . for this purpose
, we use recent luminosity functions ( lfs ) and spectral models of seyferts and fsrqs in literature . since a quantitative estimate based only on the comptel results is not easy due to the small number of detected sources , we adopt recent x - ray or gamma - ray luminosity functions and spectral models of seyferts and fsrqs .
this paper is organized as follows .
mev gamma - ray emission models of seyferts and fsrqs are described in section [ sec : sed ] .
lfs of seyferts and fsrqs are presented in section [ sec : lf ] . in section [ sec : count ] , results of the expected source counts and redshift distributions are presented .
discussions and conclusions are given in section [ sec : dis_con ] . throughout this paper
, we adopt the standard cosmological parameters of @xmath8 .
the x - ray spectra of seyfert are phenomenologically explained by a combination of following components : a primary power - law continuum with a cutoff at @xmath50.3 mev in the form of @xmath9 , absorption by surrounding gas , emission lines , a reflection component , and a soft excess of emission at @xmath10 kev ( e.g. * ? ? ?
relative fractions of these components vary with sources .
physically comptonization of disk photons in a corona above the accretion disk generate the primary power - law continuum ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the temperature of the corona roughly determines the position of the spectral cutoff and the photon index of the intrinsic continuum together with the optical depth ( see e.g. * ? ? ?
the compton reprocessed emission and bound - free absorption of the primary continuum by surrounding cold matter generate the reflection component @xcite .
, u03 seyfert spectral model with a thermal cutoff at 0.3 mev with @xmath11 , ( * ? ? ?
* i08 ) seyfert spectral model with a non - thermal tail in the mev band with @xmath12 , and ( * ? ? ?
* a09 ) fsrq spectral model having a peak at 1 mev .
the shape is assumed to be independent of luminosities for these models .
[ fig : sed_1 ] , width=302 ] although coronae with temperature of @xmath13 mev do not produce significant mev gamma - ray emission , some spectral models predict a power - law tail after the thermal cutoffs ( e.g. i08 ) .
if the corona is composed of thermal and non - thermal populations , a mev tail will appear after the cut - off .
such non - thermal electrons may exist if magnetic reconnection heats the corona @xcite as in the solar flares ( e.g. * ? ? ? * ) and
earth s magnetotail @xcite .
the mev background can be explained by the seyferts explaining the cxb which have non - thermal electrons in the coronae having @xmath14% of the total electron energy ( i08 ) .
observationally , the cutoff energy of seyferts are constrained at @xmath15 mev @xcite . the oriented scintillation spectroscopy experiment ( _ osse _ ) clearly detected emission up to 0.5 mev in the spectrum of the brightest seyfert ngc 4151 @xcite .
this measurement constrained non - thermal fraction to be @xmath1615% @xcite .
interestingly , future radio observations are able to probe those non - thermal tail through synchrotron emission @xcite . in this paper ,
as shown in figure [ fig : sed_1 ] , two primary spectral models are applied for seyferts as in @xcite .
one is thermal spectral model having a power - law continuum with a cutoff ( see e.g. * ? ? ?
* hereinafter u03 ) .
we adopt @xmath17 and @xmath18 mev .
the other is thermal plus non - thermal spectral model ( see i08 for details ) which could explain the mev background spectrum .
we adopt the same parameters as in i08 with the non - thermal photon index of 2.8 , but setting the cutoff at 0.3 mev . for the compton reflection component
, we use a compton reflection model @xcite , assuming a solid angle of @xmath19 , an inclination angle of @xmath20 , and solar abundance for all elements .
to calculate absorbed spectra , we use an absorption model called `` wabs '' developed for the xspec package . , @xmath21 , @xmath22 , and @xmath23 , respectively , based on @xcite .
[ fig : sed_2 ] , width=302 ] multi - wavelength observations allowed us to study overall spectra of blazars from radio to gamma - ray .
the observed blazar spectra consists of two broadband emission components @xcite .
one is synchrotron radiation component and the other is inverse compton ( ic ) component in which electrons scatter internal synchrotron radiation ( see e.g. * ? ? ? * ) or external radiation ( see e.g. * ? ? ?
* ; * ? ? ?
blazars are divided into two categories by their optical spectra . those are bl lacertae objects ( bl lacs ) and fsrqs . in the case of bl lacs
, the highest observable energy part of the synchrotron emission appears in the x - ray band .
x - ray spectra of bl lacs show soft spectra with a photon index of @xmath24 . in the case of fsrqs
, the lowest observable energy part of the ic emission appears in the x - ray band showing harder spectra .
although bl lacs are not expected to the dominant contributors of the mev background , fsrqs may make up the whole mev background ( a09 ) .
hence , in this paper , we do not consider bl lacs but only fsrqs for blazars .
we consider two spectral energy distribution ( sed ) models for fsrqs .
one is based on the _ swift_-bat fsrqs ( a09 ) , while the other is based on the _
fermi_-lat fsrqs ( * ? ? ?
* hereinafter a12 ) . following a09 , the sed is given by a double power - law model : @xmath25^{-1},\ ] ] where we set @xmath26 mev , @xmath27 , and @xmath28 following a09 ( see figure [ fig : sed_1 ] )
. the average photon index of the _ swift_-bat fsrqs is @xmath29 at 1555 kev ( a09 ) .
although theoretical models predict a spectral turnover at the mev band for fsrqs ( see e.g. * ? ? ?
* ) , @xmath30 and @xmath31 in the mev band are observationally unconstrained .
hence , these parameters are set fsrqs to make up the mev background . here , the latest catalog of the _ fermi_-lat sources reported detection of 467 fsrqs .
the mean of their photon index above 0.1 gev of those gamma - ray bright fsrqs is typically @xmath32 @xcite , which is harder than that of the mev background . combining with the _ fermi_-lat and _ swift_-bat detected fsrqs , the average sed is modelled in fig .
9 of a12 which are shown in figure [ fig : sed_2 ] .
we also adopt these seds as the other fsrq template . however , as discussed in the next section , the whole mev background is not be able to be explained by the a12 fsrq model .
cccc & @xcite & @xcite & @xcite + & seyfert & fsrq & fsrq + & 2 - 10 kev & 15 - 55 kev & 0.1 - 100 gev + @xmath33@xmath34 & @xmath35 & @xmath36 & @xmath37 + @xmath38@xmath39@xmath40 & @xmath41 & @xmath42 & @xmath43 + 48.0 + @xmath44 & @xmath45 & - & @xmath46 + @xmath47 & @xmath48 & @xmath49 & @xmath50 + @xmath51 & @xmath52 & - & @xmath53 + @xmath38@xmath54@xmath40 & @xmath55 & - & - + @xmath56 & @xmath57 & - & @xmath58 + @xmath59 & @xmath60 & @xmath61 & @xmath62 + @xmath63 & @xmath64 & @xmath65 & @xmath66 + @xmath67 & - & - & @xmath68 + @xmath69 & - & - & @xmath70 + + + to obtain the expected source counts and redshift distributions of agns in the mev gamma - ray band , a lf is required .
we briefly review lfs of seyferts and fsrqs in this section .
the cosmological evolution of seyferts is investigated by various x - ray surveys ( see e.g. u03 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
based on these surveys , the observed x - ray luminosity functions ( xlfs ) are known to be well represented by luminosity - dependent density evolution ( ldde ) models ( see e.g. * ? ? ?
* and references therein ) . in this study
, we follow the u03 ldde xlf at 2 - 10 kev for the seyfert models , since the i08 seyfert model is based on u03 .
we note that u03 modelled the evolution of x - ray emitting agns .
the comoving number density @xmath71 in the ldde is @xmath72 where @xmath73 is the x - ray luminosity at 2 - 10 kev before the absorption , @xmath74 is the redshift , and @xmath75 is the neutral hydrogen column density .
@xmath76 is the agn xlf at present day characterized by the faint - end slope index @xmath77 , the bright - end slope index @xmath47 , and the break luminosity @xmath39 , as : @xmath78^{-1 } \ , \ ] ] where @xmath33 is the normalization parameter having a dimension of volume@xmath79 .
the density evolution function @xmath80 is given as : @xmath81 where @xmath82 is the redshift of evolutionary peak , given as @xmath83 the function @xmath84 is the distribution of absorption column density given in the following form in the xlf @xcite : @xmath85 where @xmath861.7 and @xmath87)\},$ ] for which @xmath88 .
we note the integration of @xmath84 between @xmath89 is unity .
the parameters obtained by the fit to the observed data of x - ray agns in u03 are shown in table [ xlf - parameters ] .
we set the minimum and maximum of the x - ray luminosity as @xmath90 and @xmath91 erg s@xmath79 , respectively , the same as in u03 .
the redshift range is set to be @xmath92 , otherwise noted . to explain the cxb , the fraction of the compton thick agns between @xmath93 is set to be the same as that of the population at @xmath94 following u03 .
recently , @xcite suggested another evolution model for agns , the luminosity and density evolution ( lade ) model .
although the distribution of absorption column density is not available for the lade , we can test that model assuming the same absorption column density distribution as in u03 .
the overall source counts in the mev band significantly decrease by a factor of @xmath95 at the flux range of @xmath96@xmath0 .
however , the latest study of x - ray agn evolution @xcite found that the agn xlf is not well described with the lade model but with ldde model .
when we adopt the latest ldde model @xcite , the seyfert source counts in the mev band decrease by a factor of @xmath97% at the flux range of @xmath96@xmath0 .
as the i08 seyfert model is based on u03 , we follow the u03 ldde model in this paper . in the gev gamma - ray band ,
a12 and @xcite have recently shown that the gamma - ray luminosity functions ( glfs ) of fsrqs and bl lacs can be described by ldde models based on the _
fermi_lat fsrqs and bl lacs , which was indicated before the launch of _ fermi _
( e.g. * ? ? ?
* ; * ? ? ?
* ) . with the a12 glf model
, fsrqs can explain up to @xmath2% of the mev background . in the x - ray band ,
a09 have investigated the xlf of fsrqs using the 3-year _
swift_bat agn survey data including 26 fsrqs .
a pure luminosity evolution ( ple ) model successfully reproduced the observed distribution of x - ray fsrqs . assuming this ple model and the spectral model in .
2.2 , fsrqs explained the whole mev background .
these two recent fsrq xlf and glf models of a09 and a12 are adopted in this paper .
the fsrq xlf at 1555 kev by a09 is given in the form of the ple .
the comoving number densities @xmath71 in the ple is given as : @xmath98 where @xmath73 is the x - ray luminosity at 15 - 55 kev here .
the local xlf is described as : @xmath99 the evolution factor @xmath100 is independent of luminosity given by : @xmath101 the values of the parameters for the a09 xlf are shown in table [ xlf - parameters ] . the minimum luminosity is also assumed to evolve with redshift as @xmath102 where @xmath103 is the minimum luminosity at @xmath104 .
we set @xmath105 , @xmath106=10@xmath107 erg s@xmath79 , @xmath108 , and @xmath109 ( a09 ) , otherwise noted . on the contrary ,
the a12 fsrq glf is given in the form of ldde model .
the comoving number densities @xmath110 in the ldde is given as : @xmath111 where @xmath112 is the gamma - ray luminosity at 0.1100 gev and @xmath113 is the intrinsic photon index between 0.1100 gev .
@xmath114 is the fsrq glf at present .
this is characterized as : @xmath115^{-1}.\ ] ] the density evolution function @xmath116 is given by : @xmath117^{-1}\ ] ] where @xmath82 is given as @xmath118 the function @xmath119 describes the distribution of photon index , which is given by a gaussian form in the glf ( a12 ) : @xmath120,\ ] ] where @xmath67 and @xmath69 are the mean and the dispersion of the gaussian distribution .
however , we adopt the average sed of fsrqs ( see fig . [ fig : sed_2 ] ) not power - law seds .
therefore , we include the integrated @xmath119 over @xmath113 between 1.8 and 3.0 which is independent of luminosity and redshift . the parameters for the a12 glf are shown in table [ xlf - parameters ] .
the limits of gamma - ray luminosities , redshifts , and photon indices are set as @xmath121=10@xmath122 erg s@xmath79 , @xmath121=10@xmath123 erg s@xmath79 , @xmath124=0.01 , and @xmath125=6 , the same as in a12 , otherwise noted .
the cumulative source counts in the entire sky at a sensitivity of @xmath126 , @xmath127 , are given by integrating the luminosity function where the luminosity range is set to be @xmath128 .
@xmath129 represents the sensitivity in a given energy range between @xmath130 and @xmath131 in the unit of @xmath132}$ ] and @xmath133 is the minimum detectable luminosity with a given sensitivity @xmath129 and a redshift @xmath74 .
[ fig : count ] shows the cumulative source counts in the entire sky as a function of sensitivity limit at a given energy band in the entire sky predicted based on u03 for seyferts with the thermal cutoff , i08 for seyferts with the non - thermal tail , a09 for _ swift_bat fsrqs , and a12 for _ fermi_lat fsrqs .
the surface number density is given by dividing the number by @xmath134 steradian .
the energy bands are divided into five as 0.31 mev , 13 mev , 310 mev , 1030 mev , and 30100 mev as indicated in panels .
hereinafter , we investigate required sensitivities for detection of several hundred sources .
several hundred samples may allow us to study cosmological evolution of a population ( see u03 for x - ray agns , a12 for gamma - ray fsrqs , * ? ? ? * for gamma - ray bl lacs ) .
we also consider the sensitivity limit of @xmath3 at the mev band which is an order of magnitude better than the sensitivity of the comptel and almost the same as the expected sensitivity of the next generation mev instruments such as astrogam .
based on the u03 seyfert model , @xmath135 seyferts at 0.31 mev with the sensitivity limit of @xmath3 are expected , which would be useful to investigate the position of the thermal cutoff energies of seyferts .
once we can achieve the sensitivity limit of @xmath0 , @xmath136 seyferts at 0.31 mev .
however , only @xmath7 seyferts are expected at 13 mev .
as the u03 seyfert model relies on the thermal cutoff spectral model at 0.3 mev , we can not expect any detection of u03-type seyferts above several mev .
if the non - thermal tails of seyferts make up the mev background ( i08 ) , more seyferts at the mev band would be expected .
based on the i08 seyfert model , we can expect @xmath137 , @xmath7 , @xmath138 , and a few seyferts at 0.31 , 13 , 310 , and 1030 mev , respectively , with the sensitivity limit of @xmath3 . by achieving the sensitivity limit of @xmath0
, we can expect a reasonable number of detections as @xmath139 , @xmath140 , @xmath141 , @xmath97 , and @xmath7 seyferts at 0.31 , 13 , 310 , 1030 , and 30100 mev , respectively .
therefore , the sensitivity limit of @xmath142 is required to detect several hundred seyferts even if seyferts dominate the mev background radiation .
if the origin of the mev background is fsrqs ( a09 ) , we can expect @xmath143 , @xmath144 , @xmath2 , @xmath7 , and @xmath138 fsrqs at 0.31 , 13 , 310 , 1030 , and 30100 mev , respectively , with the sensitivity limit of @xmath3 .
once the sensitivity limit of @xmath145 is achieved , @xmath146 , @xmath147 , @xmath148 , @xmath143 , and @xmath97 fsrqs at 0.31 , 13 , 310 , 1030 , and 30100 mev will be detected , respectively , because of strong positive evolution . however , based on the latest a12 fsrq glf , with which the whole mev background can not be explained solely by fsrqs , the expected number will be @xmath149 , @xmath148 , @xmath150 , @xmath151 , and @xmath152 fsrqs at 0.31 , 13 , 310 , 1030 , and 30100 mev , respectively , with the sensitivity limit of @xmath3 . moreover , at the sensitivity limit of the comptel , @xmath4 , about 10 fsrqs are expected based on the a12 glf , which is consistent with the number of the comptel blazars . the difference between fsrq models mainly comes from the different evolutionary history . when the sensitivity limit of @xmath0 is achieved in the lowest energy band , expected detection rate for seyferts and fsrqs will be one seyfert per 2228 deg@xmath153 and one fsrq per 440 deg@xmath153 , respectively , depending on assumed models . to avoid the source confusion , we may need angular resolutions of several degrees .
astrogam will achieve the fwhm of the angular resolution measure of @xmath14 deg at 0.3 mev .
we note that comptel had those of @xmath154 deg at 0.5 mev @xcite .
it is also important to know how distant agns can be observed by future missions . since fsrqs have a spectral peak at the mev band , future mev missions will be able to detect distant fsrqs . in the gev band , the most distant blazar confirmed by _
fermi _ is pks 0537 - 286 at @xmath155 at @xmath156 mev @xcite . @xcite and @xcite have reported one egret blazar candidate at @xmath157 , but it has not been confirmed so far by _
it is also expected that _ fermi _ may eventually detect blazars at @xmath158 @xcite . @xcite
have reported a candidate _
gamma - ray blazar at @xmath159
. figure .
[ fig : z_1e-11 ] shows the cumulative redshift distribution @xmath160 as a function of redshift at a given energy band in the entire sky assuming the sensitivity limit of @xmath3 predicted based on u03 seyfert model , i08 seyfert model , a09 fsrq model , and a12 fsrq model . here , we set @xmath161 as the most distant known quasar is at @xmath162 @xcite .
the energy bands were divided into five as 0.31 mev , 13 mev , 310 mev , 1030 mev , and 30100 mev as indicated in panels .
[ fig : z_1e-12 ] shows the same but assuming the sensitivity limit of @xmath0 . for the seyfert models , the detectable seyferts will be only local seyferts .
even if the sensitivity limit of @xmath0 is achieved , the most distant seyfert will be at @xmath163 .
on the other hand , for the fsrq models , we can expect @xmath1644 fsrqs with the sensitivity limit of @xmath3 .
once the sensitivity limit of @xmath0 is achieved , @xmath165 fsrqs would be detectable even at the mev band .
in this paper , we studied the expected number counts and redshift distributions of agns ( seyferts and fsrqs ) for future mev missions based on recent agn lfs and spectral models as seyferts ( i08 ) and fsrqs ( a09 ) are discussed as the plausible origins of the cosmic mev gamma - ray background . for seyferts , we assume two primary spectral models . we considered a thermal spectral model with a cutoff at 0.3 mev ( u03 ) and a thermal plus non - thermal spectral model ( i08 ) .
we adopt the ( u03 ) ldde xlf for both seyfert models . for the thermal plus non - thermal spectral model ( i08 )
, a non - thermal component appears after the thermal cutoff at 0.3 mev with the photon index of 2.8 to explain the mev background up to a few tens of mev . for fsrqs
, we adopt two models .
one is based on the _ swift_bat detected fsrqs ( a09 ) and the other is based on the _ fermi_lat detected fsrqs ( a12 ) .
the a09 fsrq model can explain the entire mev background solely by fsrqs , while the a12 fsrq model makes up up to @xmath2% of the mev background by fsrqs .
since a thermal cutoff exists in spectra in the u03 seyfert model , we can not expect any detections at @xmath166 mev even with the sensitivity limit of @xmath0 .
in contrast , if the origin of the mev background is non - thermal tail from seyferts ( i08 ) , we can expect several hundred seyferts at the mev band with the sensitivity limit of @xmath0 .
since each seyfert is faint , we can detect only nearby ( @xmath167 ) seyferts even with the sensitivity limit of @xmath0 .
if fsrqs make up the whole mev background ( a09 ) , the sensitivity of @xmath1 is needed to detect several hundreds of fsrqs at the mev gamma - ray band . however , based on the latest fsrq glf ( a12 ) , the sensitivity limit of @xmath3 , which is almost the same as the expected sensitivity of the next generation mev telescopes , would be enough to detect several hundreds of fsrqs .
the difference between the two fsrq models comes from the different evolutionary history .
future mev observational windows will range about three orders of magnitude in energy . with that wide energy range , the dominant population of the cosmic mev gamma - ray background radiation would change with the energy . furthermore ,
as there are uncertainties of luminosity functions and spectral models , the sensitivity of several times of @xmath0 would be desirable to detect several hundred agns .
it is important to know the origin of the mev background because the expected source counts depends on it . to observationally unveil the origin of the mev background , we need to resolve the sky into point sources as in soft x - ray
, the expected sensitivity of the next generation mev instruments such as astrogam is @xmath168 . with this sensitivity ,
the expected resolved fraction will be a few percent of the total background in the i08 seyfert scenario , while it will be @xmath55% in the a09 fsrq scenario .
thus , the large fraction of the mev sky will not be resolved even with the next generation instruments . however , as about one hundred to several hundred fsrqs are expected , we will be able to obtain typical mev spectra and cosmological evolution of fsrqs and theoretically estimate their contribution to the mev background more robustly . on the other hand , it would be hard for seyferts since we can expect only about ten sources .
the angular power spectrum of the mev sky will provide a unique opportunity to unveil the origin of the mev sky @xcite .
the sgd onboard _ astro - h _ will be able to probe it up to 0.6 mev .
although the angular power spectrum with the sensitivity of the sgd is dominated by the poisson term @xmath169 , a significant difference of the @xmath169 of seyferts @xcite and fsrqs @xcite is expected .
this is because fsrqs are brighter but fewer than seyferts .
anisotropies of the mev sky would allow us to understand the origin of the mev background and be useful to design future missions .
the authors would like to thank francesco massaro for careful reading and constructive comments .
the authors also thank yasushi fukazawa and kazuhiro nakazawa for useful comments .
acknowledges support by the jaxa international top young fellowship . , k. , hartmann , d. h. , leising , m. d. , the , l. , share , g. h. , & kinzer , r. l. 1997 , in american institute of physics conference series , vol .
410 , proceedings of the fourth compton symposium , ed . c. d. dermer , m. s. strickman , & j. d. kurfess , 12231227 | while the x - ray , gev gamma - ray , and tev gamma - ray skies have been extensively studied , the mev gamma - ray sky is not well investigated after the imaging compton telescope ( comptel ) scanned the sky about two decades ago . in this paper , we investigate prospects for active galactic nuclei population studies with future mev gamma - ray missions using recent spectral models and luminosity functions of seyferts and flat spectrum radio quasars ( fsrqs ) .
both of them are plausible candidates as the origins of the cosmic mev gamma - ray background .
if the cosmic mev gamma - ray background radiation is dominated by non - thermal emission from seyferts , the sensitivity of @xmath0 is required to detect several hundred seyferts in the entire sky .
if fsrqs make up the cosmic mev gamma - ray background , the sensitivity of @xmath1 is required to detect several hundred fsrqs following the recent fsrq x - ray luminosity function . however , based on the latest fsrq gamma - ray luminosity function , with which fsrqs can explain up to @xmath2% of the mev background , we can expect several hundred fsrqs even with the sensitivity of @xmath3 which is almost the same as the sensitivity goal of the next generation mev telescopes . |
it is evident that quantum information processes such as quantum teleportation@xcite , quantum cryptography@xcite , and quantum computer@xcite will play crucial role in the future technology . in this reason , much attention is paid , recently , to the quantum entanglement@xcite because it is regarded as a genuine physical resource for the quantum information processing .
although research into the quantum entanglement has a long history@xcite , the study of its properties in the relativistic setting was initiated recently@xcite .
the main issue in this subject is to understand how a given entanglement is changed in the inertial-@xcite and non - inertial-@xcite frames . in the non - inertial frame degradation of the entanglement
occurs , which is related to the well - known unruh effect@xcite .
more recently , moreover , the change of the entanglement in the black hole background was also examined@xcite .
on the other hand , braneworld scenario@xcite , one of the modern cosmology , assumes that our @xmath2 spacetime universe is embedded in higher - dimensional world . in this reason , the absorption and emission properties of the higher - dimensional black holes
were investigated few years ago@xcite .
the purpose of this paper is to examine the degradation of the bipartite entanglement in the presence of the @xmath0-dimensional schwarzschild black hole . in order to explore the degradation we assume that alice and rob share the maximally entangled state initially . after sharing rob moves to the near - horizon region while alice is free falling into the black hole . in this situation
we will compute the entanglement by adopting the negativity@xcite as an entanglement measure .
our main focus in this paper is to investigate the effect of the extra dimensions @xmath1 in the entanglement degradation . of course ,
the result of ref .
@xcite is reproduced when @xmath3 .
the paper is organized as follows . in section
ii we review the spacetime geometry of the @xmath0-dimensional schwarzschild black hole . for later convenience
we compare the @xmath0-dimensional schwarzschild spacetime with the rindler spacetime in this section .
this comparison enables us to transform our problem into the degradation of entanglement in the non - inertial frame , which was studied in ref .
@xcite . in section
iii we discuss on the @xmath1-dependence of the entanglement .
the correlation between alice and rob is generally degraded in the presence of the black hole .
it is found that the degradation becomes weaker and weaker with increasing @xmath1 .
for the case between rob and antirob the quantum correlation , contrary to the physically accessible correlation , decreases with increasing @xmath1 . in section
iv a brief conclusion is given . in appendix a explicit computation of negativity is performed .
the higher - dimensional black hole solutions of the einstein field equation were discussed in detail in ref.@xcite .
the explicit expression of the @xmath0-dimensional schwarzschild black hole solution in terms of the usual schwarzschild coordinates is in the following : @xmath4 where @xmath5 the horizon radius @xmath6 is related to the black hole mass @xmath7 as following : @xmath8 where @xmath9 is a @xmath0-dimensional planck mass and @xmath10 is a newton constant .
because of the symmetry of the problem we restrict our attention to the temporal and radial coordinates .
thus , the part of the metric we will analyze is @xmath11 we can write the line - element @xmath12 in terms of the proper time @xmath13 of an observer located in @xmath14 as follows : @xmath15 where @xmath16 and @xmath17 .
now , we define a new spatial coordinate @xmath18 as follows : @xmath19 then , the profile function @xmath20 can be written as @xmath21 where @xmath22 is a surface gravity defined by @xmath23 in terms of @xmath18 it is easy to show that the line - element @xmath24 becomes @xmath25^{(1-n ) / ( 1+n ) } dz^2.\ ] ] in the near - horizon region , _ i.e. _ @xmath26 , eq . ( [ metric4 ] )
reduces to @xmath27 this is just the metric of the rindler space with the acceleration parameter @xmath28 . in order to discuss on the physical states we define the kruskal coordinates
@xmath29 where the `` tortoise '' coordinate @xmath30 is defined by @xmath31 in terms of the kruskal coordinates the line - element @xmath24 becomes @xmath32 since @xmath30 reduces to @xmath33 at the near - horizon region , we can re - express @xmath24 in this region as @xmath34 as authors in ref .
@xcite suggested , there are three regions in this background , in which we can define the different time - like killing vectors .
thus , one should define the physical vacuum in each region .
first vacuum is known as the hartle - hawking vacuum @xmath35 .
this is defined by the time - like killing vector which is proportional to @xmath36 .
second and third vacua are the boulware vacuum@xcite @xmath37 and anti - boulware vacuum @xmath38 , which are defined by the time - like killing vectors @xmath39 and @xmath40 , respectively .
since the metric ( [ metric5 ] ) reduces to the rindler space metric with the acceleration @xmath28 in the near - horizon region , one can derive the interrelations between these vacua using the unruh effect@xcite .
the relations beyond the single - mode approximation were derived in ref .
corresponding @xmath35 to minkowski vacuum and , @xmath37 and @xmath38 to the vacua in the left- and right - wedges of rindler space , one can derive the following relations : @xmath41\end{aligned}\ ] ] where @xmath42 and @xmath43 in eq .
( [ inter1 ] ) @xmath44 means @xmath45-particle states of energy @xmath46 , which is constructed by operating the creation operator @xmath45 times to the vacuum @xmath47 with @xmath48 . in next section
we will use eq .
( [ inter1 ] ) to analyze the bosonic entanglement degradation in the presence of the black hole ( [ schwarz1 ] ) .
let us consider a situation that alice and rob initially share the maximally entangled state@xmath49 when both are free falling into the black hole . since @xmath36 is proportional to the time - like killing vector for the free falling observer , eq .
( [ maximal ] ) is understood to be written as a hartle - hawking basis . for simplicity ,
the energy parameters @xmath50 and @xmath51 for alice and rob are ignored in eq .
( [ maximal ] ) .
now , we assume that after sharing @xmath52 , rob moves to the near - horizon region , _
i.e. _ @xmath53 .
thus , we restrict ourselves into @xmath54 throughout this paper . therefore , eq . ( [ acceleration1 ] ) implies @xmath55 where @xmath56 . in this assumption eq .
( [ inter1 ] ) implies that @xmath52 is changed into @xmath57.\ ] ] in this paper our main interest is to examine the effect of the extra dimension @xmath1 in the entanglement degradation .
thus , we will choose @xmath58 for simplicity . furthermore ,
as shown in ref.@xcite , eq .
( [ inter1 ] ) with @xmath58 case corresponds to the unruh transformation within the single - mode approximation .
now , let us discuss on the entanglement between alice and rob .
in order to examine the entanglement in the presence of the @xmath0-dimensional black hole , we should compute the bipartite entanglement of a quantum state @xmath59 , where the subscript @xmath60 stands for alice and rob .
although there are a lot of entanglement measures which quantify the bipartite entanglement such as entanglement of formation@xcite , concurrence@xcite , and relative entropy of entanglement@xcite , most of them are very hard to compute mainly due to that fact that @xmath59 is a state in qudit system .
therefore , we choose a negativity@xcite in this paper for tractable computation .
the negativity @xmath61 of @xmath59 is defined as @xmath62 where @xmath63 are the eigenvalues of @xmath64 and @xmath65 is a partial transposition with respect to the party @xmath66 .
it is worthwhile noting that the negativity @xmath61 in the absence of the black hole is @xmath67 .
-dependence of @xmath61 when @xmath68 .
we fix @xmath46 as @xmath69 . in ( b ) we plot the @xmath46-dependence of @xmath70 when @xmath68 .
we fix @xmath71 as @xmath72 .
, title="fig:",height=245 ] -dependence of @xmath61 when @xmath68 .
we fix @xmath46 as @xmath69 . in ( b )
we plot the @xmath46-dependence of @xmath70 when @xmath68 .
we fix @xmath71 as @xmath72 . , title="fig:",height=245 ] the computation of @xmath61 is described in appendix a. in fig .
1 we plot @xmath61 with varying the number of extra dimensions as @xmath68 . in fig .
1(a ) we plot the @xmath71-dependence of @xmath61 with choosing @xmath69 . as this figure exhibits , @xmath61 is less than @xmath73 , so that as expected , the degradation of the entanglement occurs .
the rate of the degradation decreases with increasing @xmath71 , which means that rob goes away from the black hole . when rob approaches to the horizon , _
i.e. _ @xmath74 , all bipartite entanglement vanishes .
another remarkable fact this figure shows is that the degradation becomes weaker with increasing @xmath1 .
this fact can be explained as follows . in @xmath0 dimension
the newtonian gravitational force is proportional to @xmath75 .
therefore , it goes weaker and weaker with increasing @xmath1 .
thus , the effective acceleration rob needs to remain outside the black hole becomes smaller with increasing @xmath1 .
it causes the weaker degradation of the entanglement between alice and rob . in fig .
1(b ) we plot the @xmath46-dependence of @xmath61 with choosing @xmath72 .
this figure also shows that the degradation becomes weaker with increasing @xmath1 .
the reason for this fact can be explained same way .
the computation of negativity between alice and antirob is given in appendix a. it is shown that no entanglement is created between these parties when @xmath58 , which corresponds to the unruh transformation within the single - mode approximation .
-dependence of @xmath76 when @xmath77 .
we fix @xmath46 as @xmath69 . in ( b )
we plot the @xmath46-dependence of @xmath78 when @xmath77 .
we fix @xmath71 as @xmath72 .
, title="fig:",height=245 ] -dependence of @xmath76 when @xmath77 .
we fix @xmath46 as @xmath69 . in ( b )
we plot the @xmath46-dependence of @xmath78 when @xmath77 .
we fix @xmath71 as @xmath72 . ,
title="fig:",height=245 ] in appendix a the negativity @xmath79 is computed . in fig .
2 we plot @xmath79 with varying the number of extra dimensions as @xmath77 . in fig .
2(a ) we plot the @xmath71-dependence of @xmath76 with choosing @xmath69 . in fig .
2(b ) we also plot the @xmath46dependence of @xmath76 with choosing @xmath72 .
contrary to the quantum correlation between alice and rob the quantum entanglement between rob and antirob decreases with increasing @xmath1 .
we do not know why @xmath1-dependence of @xmath76 is different from that of @xmath61 .
probably , this is due to the fact that this entanglement is unphysical because rob is causally disconnected fro antirob .
thus , this quantum correlation is purely theoretical and is useless for the quantum information task .
in this short paper we compute the various bipartite quantum correlations in the presence of the @xmath0-dimensional schwarzschild black hole .
in particular , we focus on the @xmath1-dependence of the bosonic entanglements . for the case between alice and
rob the quantum correlation is degraded as expected .
the degradation rate decreases with increasing @xmath1 .
however , it is found that there is no creation of quantum correlation between alice and antirob .
for the case between rob and antirob the quantum entanglement is created although they are separated in the causally disconnected regions .
the entanglement is found to decrease with decreasing @xmath1 . in this paper
we performed the calculation under the condition of @xmath80 .
if we relax this condition , various correlations may exhibit different behavior . in this case , however , we can not use the rindler - anology of the schwarzschild background .
thus we have to re - derive the equations corresponding to eq .
( [ inter1 ] ) in this case by computing appropriate bogoliubov coefficients .
however , this generalization might be difficult problem mainly due to the difficulty in the analytic computation of the bogoliubov coefficients .
it seems to be of interest to extend this paper to the higher - dimensional rotating black hole case@xcite .
the condition for occurring the superradiance in the higher - dimensional rotating black holes was derived in ref .
it seems to be highly interesting to examine the effect of the superradiance to the entanglement degradation .
99 c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres and w. k. wootters , _ teleporting an unknown quantum state via dual classical and einstein - podolsky - rosen channels _ , phys .
* 70 * ( 1993 ) 1895 . c. h. bennett and g. brassard , _
quantum cryptography : public key distribution and coin testing _ , in proceedings of the ieee international conference on computer , systems , and signal processing , bangalore , india ( ieee , new york , 1984 ) , pp .
175 - 179 ; a. k. ekert , _ quantum cryptography based on bell s theorem _ ,
* 67 * ( 1991 ) 661 ; c. a. fuchs , n. gisin , r. b. griffiths , c. s. niu and a. peres , _ optimal eavesdropping in quantum cryptography . i. information bound and optimal strategy _ , phys
. rev . * a 56 * ( 1997 ) 1163 [ quant - ph/9701039 ] .
g. vidal , phys .
lett . _ efficient classical simulation of slightly entangled quantum computations _ , * 91 * ( 2003 ) 147902 [ quant - ph/0301063 ] ; m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _
( cambridge university press , cambridge , england , 2000 ) .
r. horodecki , p. horodecki , m. horodecki , and k. horodecki , _ quantum entanglement _
* 81 * ( 2009 ) 865 [ quant - ph/0702225 ] and references therein .
a. einstein , b. podolsky and n. rosen , _ can quantum - mechanical description of physical reality be considered complete ? _ , phys .
* a47 * ( 1935 ) 777 .
e. schrdinger , _ die gegenwrtige situation in der quantenmechanik _ , naturwissenschaften , * 23 * ( 1935 ) 807 .
a. peres and d. r. terno , _ quantum information and relativity theory _ , rev .
* 76 * ( 2004 ) 93 [ quant - ph/0212023 ] .
a. peres and d. r. terno , _ quantum information and special relativity _ , int . j. quant
* 1 * ( 2003 ) 225 [ quant - ph/0301065 ] .
a. peres , _ quantum information and general relativity _ , fortsch . phys .
* 52 * ( 2004 ) 1052 [ quant - ph/0405127 ] .
m. czachor , _ einstein - podolsky - rosen - bohm experiment with relativistic massive particles _ ,
phys . rev . * a 55 * ( 1997 ) 72 [ quant - ph/9609022 ] .
a. peres , p. f. scudo and r. terno , _ quantum entropy and special relativity _ , phys .
* 88 * ( 2002 ) 230402 [ quant - ph/0203033 ] .
p. m. alsing and g. j. milburn , _ lorentz invariance of entanglement _ , quantum inf .
* 2 * ( 2002 ) 487 [ quant - ph/0203051 ] .
r. m. gingrich and c. adami , _ quantum entanglement of moving bodies _
* 89 * ( 2002 ) 270402 [ quant - ph/0205179 ] .
i. fuentes - schuller and r. b. mann , _ alice falls into a black hole : entanglement in noninertial frames _ , phys .
* 95 * ( 2005 ) 120404 [ quant - ph/0410172 ] .
y. ling , s. he , w. qiu and h. zhang , _ quantum entanglement of electromagnetic field in non - inertial reference frames _ , j. phys : math .
a40 * ( 2007 ) 9025 [ quant - ph/0608029 ] . q. pan and j. jing , _ degradation of nonmaximal entanglement of scalar and dirac fields in noninertial frames _ , phys . rev . * a 77 * ( 2008 ) 024302 [ arxiv:08021238 ( quant - ph ) ] .
a. datta , _ quantum discord between relatively accelerated obserbers _
, phys . rev . * a 80 * ( 2009 ) 052304 [ arxiv:0905.3301 ( quant - ph ) ] .
j. wang , j. deng , and j. jing , _ classical correlation and quantum discord sharing of dirac fields in noninertial frames _ , phys . rev . * a 81 * ( 2010 ) 052120 [ arxiv:0912.4129 ( quant - ph ) ]
. m. r. hwang , d. k. park , and e. jung , _ tripartite entanglement in a noninertial frame _ , phys .
rev . * a 83 * ( 2011 ) 012111 [ arxiv:1010.6154 ( hep - th ) ] .
e. martin - martinez and i. fuentes , _ redistribution of particle and antiparticle entanglement in noninertial frames _ , phys .
rev . * a 83 * ( 2011 ) 052306 [ arxiv:1102.4759 ( quant - ph ) ] .
m. montero and e. martin - martinez , _ fermionic entanglement ambiguity in noninertial frames _ , phys . rev . * a 83 * ( 2011 ) 062323 [ arxiv:1104.2307 ( quant - ph ) ] .
m. montero , j. leon , and e. martin - martinez , _ fermionic entanglement extinction in noninertial frames _ , phys . rev . * a 84 * ( 2011 ) 042320 [ arxiv:1108.1111 ( quant - ph ) ] .
a. smith and r. b. mann , _ persistence of tripartite nonlocality for non - inertial observers _ , arxiv:1107.4633 ( quant - ph ) .
d. k. park , _ tripartite entanglement - dependence of tripartite non - locality in non - inertial frame _ , arxiv:1201.1335 ( quant - ph ) .
mi - ra hwang , eylee jung , and daekil park , _ three - tangle in non - inertial frame _ , will appear in class .
quantum grav .
[ arxiv:1203.5594 ( quant - ph ) ] .
w. g. unruh , _ notes on black - hole evaporation _ , phys
* d 14 * ( 1976 ) 870 .
n. d. birrel and p. c. w. davies , _ quantum fields in curved space _ ( cambridge university press , cambridge , england , 1982 ) .
q. pan and j. jing , _ hawking radiation , entanglement , and teleportation in the background of an asymptotically flat static black hole _ , phys .
* d 78 * ( 2008 ) 065015 [ arxiv:0809.0811 ( gr - qc ) ] .
e. martn - martnez , l. j. garay , and j. len , _ unveiling quantum entanglement degradation near a schwarzschild black hole _ , phys
d 82 * ( 2010 ) 064006 [ arxiv:1006.1394 ( quant - ph ) ] .
eylee jung , mi - ra hwang , and daekil park , _ quantum discord and quantum entanglement in the background of an asymptotically flat static black holes _ , arxiv:1206.5608 ( hep - th ) . n.
arkani - hamed , s. dimopoulos and g. dvali , _ the hierarchy problem and new dimensions at a millimeter _ , phys
b 429 * ( 1998 ) 263 [ hep - ph/9803315 ] . l. antoniadis , n. arkani - hamed , s. dimopoulos and g. dvali , _ new dimensions at a millimeter to a fermi and superstrings at a tev _ , phys . lett . * b 436 * ( 1998 ) 257 [ hep - ph/9804398 ] .
l. randall and r. sundrum , _ a large mass hierarchy from a small extra dimension _ , phys .
rev . lett .
83 ( 1999 ) 3370 [ hep - ph/9905221 ] .
l. randall and r. sundrum , _ an alternative to compactification _ , phys .
83 ( 1999 ) 4690 [ hep - th/9906064 ] . c. m. harris and p. kanti , _ hawking radiation from a @xmath0-dimensional black hole : exact results for the schwarzschild phase _
, jhep * 0310 * ( 2003 ) 014 [ hep - ph/0309054 ] .
e. jung and d. k. park , _ absorption and emission spectra of an higher - dimensional reissner - nrdstrm black hole _ ,
* b 717 * ( 2005 ) 272 [ hep - th/0502002 ] .
e. jung and d. k. park , _ bulk versus brane emissivities of photon fields : for the case of higher - dimensional schwarzschild phase _ ,
bf b 766 ( 2007 ) 269 [ hep - th/0610089 ] .
v. frolov and d. stojkovic , _ quantum radiation from a @xmath81-dimensional rotating black hole _ , phys
* d 67 * ( 2003 ) 084004 [ gr - qc/0211055 ] .
e. jung and d. k. park , _ bulk versus brane in the absorption and emission : @xmath82 rotating black hole case _ , nucl .
* b 731 * ( 2005 ) 171 [ hep - th/0506204 ] . g. vidal and r.f .
werner , _ a computable measure of entanglement _ , phys . rev . * a 65 * ( 2002 ) 032314 [ quant - ph/0102117 ] .
r. c. myers and m. j. perry , _ black holes in higher dimensional space - times _ , ann . phys .
* 172 * ( 1986 ) 304 .
d. g. boulware , _ quantum field theory in schwarzschild and rindler spaces _ , phys
d 11 * ( 1975 ) 1404 .
d. e. bruschi _
et al _ , _ unruh effect in quantum information beyond the single - mode approximation _ ,
phys . rev . * a 82 * ( 2010 ) 042332 [ arxiv:1007.4670 ( quant - ph ) ] . c. h. bennett , d. p. divincenzo , j. a. smolin and w. k. wootters , _ mixed - state entanglement and quantum error correction _ ,
* a54 * ( 1996 ) 3824 [ quant - ph/9604024 ] .
w. k. wootters , _ entanglement of formation of an arbitrary state of two qubits _ , phys .
lett . * 80 * , 2245 ( 1998 ) [ quant - ph/9709029 ] .
v. vedral , m. b. plenio , m. a. rippin and p. l. knight , _ quantifying entanglement _ , phys .
* 78 * ( 1997 ) 2275 [ quant - ph/9702027 ] .
eylee jung , sunghoon kim , d. k. park ,
_ condition for superradiance in higher - dimensional rotating black holes _ ,
* b 615 * ( 2005 ) 273 [ hep - th/0503163 ] ; _ condition for the superradiance modes in higher - dimensional rotating black holes with multiple angular momentum parameters _ , _ ibid _ , * b 619 * ( 2005 ) 347 [ hep - th/0504139 ] .
now , we assume that rob is in the region where killing vector is @xmath39 .
since , then , rob can not access the region whose time - like killing vector is @xmath40 , one should take a partial trace over @xmath83 in eq .
( [ black-1 ] ) . thus , the resulting quantum state for alice - rob system becomes a mixed state in a form @xmath84.\end{aligned}\ ] ] for @xmath58 computation of @xmath61 is straightforward from eq .
( [ state - ar ] ) .
since @xmath64 can be diagonal in terms of @xmath85 matrix if one choose the order of the basis appropriately , the eigenvalues can be easily computed , which is @xmath86 where @xmath87.\ ] ] since @xmath88 and @xmath89 , the negativity between alice and rob becomes @xmath90 the extra - dimensional dependence of @xmath61 arises from @xmath1-dependence of @xmath91 as eq . ( [ statistical-1 ] ) shows . from eq .
( [ black-1 ] ) with @xmath58 the state for alice - antirob becomes @xmath92.\end{aligned}\ ] ] since , by similar way , one can show that all eigenvalues of @xmath93 are positive , we get @xmath94 .
thus , no entanglement is created between alice and antirob . from eq .
( [ black-1 ] ) with @xmath58 the state for rob - antirob becomes @xmath95.\end{aligned}\ ] ] since @xmath96 is block - diagonal in terms of @xmath97 matrix provided that the order of the basis is selected as @xmath98 one can compute the eigenvalues in principle .
therefore , it is possible to compute @xmath76 numerically by making use of eq .
( [ negativity1 ] ) . | in this paper we compute the various bipartite quantum correlations in the presence of the @xmath0-dimensional schwarzschild black hole . in particular , we focus on the @xmath1-dependence of various bosonic bipartite entanglements . for the case between alice and rob , where the former is free falling observer and the latter is at the near - horizon region , the quantum correlation is degraded compared to the case in the absence of the black hole . the degradation rate increases with decreasing @xmath1 .
we also compute the physically inaccessible correlations .
it is found that there is no creation of quantum correlation between alice and antirob .
for the case between rob and antirob the quantum entanglement is created although they are separated in the causally disconnected regions .
it is found that contrary to the physically accessible correlation the entanglement between rob and antirob decreases with increasing @xmath1 . |
the adiabatic theorem of quantum mechanics is usually stated and proved for systems with a finite spectral gap .
mathematically it is known since at least 1998 @xcite that in a weaker sense the theorem remains valid even for systems without spectral gap as long as the spectral projections of the time - dependent hamiltonian are sufficiently regular functions of time , see also @xcite .
its validity has been shown even for resonances @xcite and for open systems @xcite . in recent years
the problem of adiabaticity for gapless systems attracted also interest in physics .
for example , in @xcite the authors consider adiabatic changes in the coupling of a scalar field . in a finite volume
the ground state of the field is separated by a gap from the excited states . however , in the thermodynamic limit the gap closes and the spectrum of the field becomes absolutely continuous .
the authors find that in the thermodynamic limit the error in the adiabatic approximation depends on the spatial dimension and other parameters controlling the spectral density of the field .
they distinguish three regimes : one where the adiabatic theorem holds with the same error estimate as in the gapped case , one where it holds with a different power law and one where it fails completely . in all cases
, however , they identify the error of the adiabatic approximation with the size of the non - adiabatic transitions . one main goal of our paper is to explain why this is , in general , not a valid identification and why the concept of super - adiabatic approximations is useful also in the gapless case . to this end
we consider a specific gapless model in the adiabatic limit , namely a massless scalar field with time dependent sources . the goal is to exhibit a number of subtleties in adiabatic theory that are known in the gapped case but often not expressed sufficiently clearly also for the case of gapless systems .
in particular we emphasize the difference between the error in the adiabatic approximation and what physically should be considered non - adiabatic transitions . to be more specific , let us briefly recall the situation in the presence of a spectral gap .
let @xmath0 be a time - dependent hamiltonian and @xmath1 an eigenvalue with spectral projection @xmath2 .
under appropriate regularity conditions on @xmath0 and the assumption that @xmath1 is separated by a gap from the rest of the spectrum of @xmath0 , the adiabatic theorem states that the solution of the schrdinger equation @xmath3 satisfies @xmath4 i.e. that any solution starting in the spectral subspace @xmath5 evolves into the subspace @xmath6 up to an error of order @xmath7 .
this error estimate , which contributes to the error of the adiabatic approximation , is optimal in the sense that generically the piece @xmath8 of the solution in the orthogonal complement of @xmath6 is really of order @xmath7 and not smaller for any finite time @xmath9 where @xmath10 .
however , if @xmath0 is constant outside of a bounded interval , say @xmath11 $ ] , and @xmath12-times continuously differentiable , then clearly @xmath13 for @xmath14 , but also @xmath15 thus the non - adiabatic transitions into the orthogonal complement of @xmath6 after a compactly supported adiabatic change of the hamiltonian are much smaller than @xmath16 .
what happens is that during the adiabatic change of the hamiltonian the solution does not exactly follow the spectral subspace @xmath6 , which we will call the adiabatic subspace in the following , but slightly tilted @xmath7-dependent subspaces ran@xmath17 , so - called superadiabatic subspaces . with respect to the superadiabatic subspaces one has @xmath18 in addition it holds that @xmath19 which together with ( [ ia3 ] ) implies the usual adiabatic estimate ( [ ia1 ] ) .
moreover , at times @xmath9 when all derivatives of @xmath0 vanish , the superadiabatic projections agree with the adiabatic projection , @xmath20 , and thus ( [ ia3 ] ) implies also ( [ ia2 ] ) . in figure [ transhist ]
the situation is plotted schematically .
$ ] and outside of this interval the adiabatic subspace @xmath2 , i.e. the instantaneous spectral subspace , agrees with the superadiabatic subspace @xmath17 .
while the non - adiabatic transitions for @xmath21 are of order @xmath22 , the norm of the piece of the solution that leaks into the orthogonal complement of @xmath2 for @xmath23 is typically much larger , namely of order @xmath7 .
the norm of the piece of the solution that leaks into the orthogonal complement of the superadiabatic subspace @xmath17 for @xmath23 remains small , i.e. of order @xmath22 , for all times.[transhist],height=151 ] in this sense we say that the error of the adiabatic approximation is of order @xmath7 , while the non - adiabatic transitions are ( at most ) of order @xmath22 . for a detailed exposition of adiabatic theory with spectral gap
we refer to @xcite and references therein . here
let us mention only that the notion `` superadiabatic '' was termed by m. berry in @xcite .
there he shows how for hamiltonians depending analytically on time the transitions between superadiabatic subspaces are even exponentially small in @xmath7 ( as it is known also from the famous landau - zener model ) and moreover , that the transitions as a function of time follow the universal shape of an error function , i.e. an integrated gaussian .
a rigorous account of @xcite is given in @xcite .
in the present paper we show within a physically relevant but relatively simple example that in the gapless case the situation is to some extend similar . before presenting our results in detail
let us introduce the model and explain some of its important features .
we consider a scalar massless field in three spatial dimensions .
the momentum space for a single boson is @xmath24 and the state space of the field is the symmetric fock space @xmath25 so @xmath26 is a sequence @xmath27 with @xmath28 a square integrable symmetric function of @xmath29 variables in @xmath30 .
for @xmath31 the inner product is @xmath32 the hamiltonian of the free field is @xmath33 and acts as @xmath34 as a multiplication operator @xmath35 is self - adjoint on its dense maximal domain @xmath36 .
moreover , @xmath35 has a unique ground state given by the fock vacuum @xmath37 .
now we add moving charges as sources to the field . for notational simplicity
we assume that all sources have the same normalized form factor @xmath38 with @xmath39 later on we will make additional assumptions on the `` charge distribution '' @xmath40 .
the sources are located at positions @xmath41 with total charge @xmath42 and form factor @xmath40 . the linear coupling to the field is given by the operator @xmath43 where we abbreviate @xmath44 . here
@xmath45 denotes the fourier transform of @xmath40 and @xmath46 is the field operator @xmath47 where @xmath48 and @xmath49 are the standard bosonic annihilation and creation operators on @xmath50 , see section [ sec : fock ] for more details . under appropriate conditions on @xmath40
the total hamiltonian @xmath51 is self - adjoint on @xmath52 and bounded from below .
if the total charge of the sources @xmath53 is zero , @xmath54 has a unique ground state @xmath55 .
the ground state @xmath56 contains so - called `` virtual bosons '' that provide a `` dressing '' of the sources .
in contrast to `` free bosons '' , the virtual bosons do not propagate .
these heuristic notions will be made more precise later on .
if the total charge is different from zero there is still a good notion of approximate ground state , at least for our purposes
. we will be interested in the dynamics of the scalar field when the sources change their locations with time .
let @xmath57 , @xmath58 be smooth , then @xmath59 describes the evolution of the scalar field in the presence of sources moving along the prescribed trajectories @xmath60 .
we consider the solutions of the schrdinger equation @xmath61 for the field in the adiabatic limit @xmath62 , which corresponds to slowly moving sources .
assume for a moment that the field is restricted to a finite box .
then the hamiltonian @xmath0 has purely discrete spectrum and the ground state is separated from the first excited state by a gap .
the adiabatic theorem implies that if the field starts in the ground state @xmath63 of @xmath64 , it remains up to a phase @xmath7-close to the ground state @xmath65 of @xmath0 at later times .
the solution is thus @xmath7-close to a state where a static cloud of virtual bosons follows the slowly moving sources and no radiation is emitted .
the error of the adiabatic approximation is of order @xmath7 .
if we pass to the superadiabatic approximation , we realize that the solution is even @xmath22-close to a state @xmath66 where an @xmath7-dependent cloud of virtual bosons follows the slowly moving sources and no radiation is emitted .
the state @xmath66 is still @xmath7-close to the static ground state @xmath65 , however , the dressing by virtual bosons depends now also on the velocities @xmath67 and higher derivatives of @xmath60 . if at some time the sources come to rest then from that time on @xmath68 , hence the field is @xmath22-close to the static ground state again . in particular
, the probability for emitting a free boson is at most of order @xmath69 .
the sources are at rest and the adiabatic approximation agrees with the true solution . while the sources are moving ,
the dressing is deformed , which leads to an error of order @xmath7 in the adiabatic approximation .
the superadiabatic approximation takes into account the deformation and the error is only of order @xmath22 . when the sources are again at rest for @xmath21 , the dressing is again the static one and the error of the adiabatic and the superadiabatic approximation is of order @xmath22
this error is due to non - adiabatic transitions that correspond to the emission of free bosons.[f2 ] , height=377 ] summarizing the above , we have that as long as there is a gap and as long as the prescribed trajectories @xmath60 are @xmath70 , non - adiabatic transitions corresponding to the emission of free bosons are at most of order @xmath22 . nonetheless , the error of the adiabatic approximation is of order @xmath7 for all times where @xmath71 because of the deformation of the cloud of virtual bosons , see also figure [ f2 ] .
this exemplifies in a simple physical example why it is important to distinguish carefully between the error of the adiabatic approximation and the non - adiabatic transitions .
it also is an example where superadiabatic subspaces have a clear physical meaning as velocity - dependent dressing of the sources .
this is of course all well known and the content of this paper is to show that this picture survives also in the case without spectral gap , i.e. in the thermodynamic limit where the box is replaced by @xmath30 . in the remainder of the introduction we provide informal statements of our main results , where for simplicity we restrict to the case of a neutral system . then there is a unitary transformation @xmath72 such that @xmath73 and thus @xmath54 has a unique ground state @xmath74 with eigenvalue @xmath75 .
the general and rigorous statements are explained in section 2 .
our first result is an adiabatic theorem without spectral gap , cf
. theorem [ thm : adapp ] . + * adiabatic theorem : * _ the solution of @xmath76 with @xmath77 satisfies for any @xmath78 @xmath79 _
thus , although the spectrum of @xmath54 is the whole half line @xmath80 and the eigenvalue @xmath75 is not separated by a gap from the continuous spectrum , the usual adiabatic approximation for the evolution of eigenstates holds with almost the same error estimate as in the gapped case .
we have even more : if one adds free bosons then their dynamics is adiabatic too in the following sense .
to each configuration @xmath81 of the sources the annihilation and creation operators of free bosons are @xmath82 and @xmath83 and thus @xmath84 where @xmath85 is just the free time evolution of a boson and @xmath86 stands either for @xmath87 or @xmath88 .
now we define the adiabatic approximation as @xmath89 we will show that it approximates the solution of ( [ schroe ] ) with initial datum @xmath90 in the same sense as in ( [ adi0 ] ) , @xmath91 if we denote by @xmath92 the projection on the sector of fock space containing exactly @xmath93 free bosons , where @xmath94 is the projection on the @xmath95-particle sector of fock space , the above result implies that for @xmath96 @xmath97 i.e. that the probability for emitting a free boson is at most of order @xmath98 . note that this means that not only the spectral subspace @xmath99 is adiabatically invariant , but also the subspaces @xmath100 , which are not spectral subspaces of the instantaneous hamiltonian @xmath101 . in order to understand what part of @xmath102 really corresponds to emission of free bosons and what part is merely a velocity - dependent deformation of the dressing
, we need to introduce the superadiabatic picture . now
not only the dressing of the sources depends on their velocities , but also the annihilation and creation operators of free bosons and the associated @xmath95-particle sectors of fock space . in lemma
[ lem : v ] we construct a velocity - dependent dressing operator @xmath103 such that the corresponding objects @xmath104 , @xmath105 and @xmath106 allow for an improved adiabatic approximation .
let @xmath107 be the superadiabatic approximation to the solution @xmath108 of ( [ schroe ] ) with initial datum @xmath109 then ( [ adiapprox ] ) can be improved in the following sense : the part of the solution @xmath108 that stays in the @xmath95-particle sector is given by @xmath110 with a much smaller error , @xmath111 however , more importantly , the superadiabatic representation can be used to compute the non - adiabatic transitions corresponding to the emission of free bosons by first order perturbation theory . in theorem
[ thm : foae ] we show that the `` non - adiabatic '' wave function @xmath112 satisfies @xmath113 here @xmath114 and @xmath115 is a coupling function depending on the acceleration of the sources . in summary @xmath116 and thus @xmath117 yields a good approximation of the true solution @xmath108 with a clear separation of a `` superadiabatic '' piece @xmath110 that improves the adiabatic approximation and a `` non - adiabatic '' piece @xmath118 that contains real non - adiabatic transitions . note that a posteriori one can compute the leading order of @xmath118 replacing all superadiabatic objects by their leading order adiabatic counterparts , @xmath119
finally one can ask for the probability of emitting a free boson , i.e. for computing @xmath120 .
however , due to the infrared problem it turns out that this has no nice asymptotics for @xmath121 and we can only show @xmath122 . but a physically more interesting question is the energy radiated through the emission of free bosons . in theorem [ thm :
rad ] we show that when starting in the dressed vacuum , i.e. @xmath123 , then @xmath124 where @xmath125 is the second derivative of the dipole moment of the sources and @xmath126 is the field energy of the free photons . taking a derivative gives the larmor formula for the radiated power of slowly moving charges , @xmath127 to come back to our main message once more , note that computing naively the energy in the piece of the solution that constitutes the error of the adiabatic approximation @xmath128 would have led to @xmath129 where @xmath130 .
as our analysis showed , this velocity - dependent term is the energy in the deformation of the dressing and being of order @xmath131 , it is much bigger than the energy emitted by radiation of free bosons , which is of order @xmath132 . in order to obtain the correct picture
, the use of superadiabatic approximations seems inevitable .
finally let us mention that a closely related problem was considered in @xcite , see also @xcite .
there semiclassical limit of non - relativistic schrdinger particles coupled to a massless scalar field was considered . in a sense
the system considered in the present paper can be considered a toy model version of the one in @xcite . however , as a consequence of the simplicity of the model the results obtained here are much stronger and at the same time the proofs are more transparent . for + * acknowledgements .
* we thank rainer nagel for pointing out to us reference @xcite .
we always assume the following properties for the parameters of the model : the charge distribution @xmath133 is a spherically symmetric schwartz - function with @xmath134 and @xmath135 for @xmath136 . as explained in the introduction , the central object that we construct is the transformation to the superadiabatic representation , a unitary dressing transformation @xmath137 in the superadiabatic representation the schrdinger equation reads @xmath138 in this representation the fock vacuum @xmath139 corresponds to the dressed ground state and the @xmath95-particle sectors of fock space correspond to states with @xmath95 free bosons . while in the introduction we formulated the statements in the original representation , it is more convenient to formulate and prove everything in the new representation after performing the dressing transformation .
the translation back to the original representation is straightforward using the definitions of @xmath140 of the introduction .
note , however , that if the total charge is different from zero , the transformation @xmath141 has no asymptotic expansion in powers of @xmath7 , not even the limit @xmath142 exists .
let @xmath143 denote the projection on the sectors of fock space with at most @xmath144 photons .
[ thm : adapp ] the dressed hamiltonian @xmath145 generates a unitary propagator @xmath146 . for any @xmath147
there are constants @xmath148 and @xmath149 such that for all @xmath150 , @xmath151 and @xmath152 @xmath153 in the norm of @xmath154 and in the norm of @xmath155 . here
@xmath156 with @xmath157 thus at leading order the time evolution is just the free evolution of the bosons and a time - dependent dynamical phase . in particular , the @xmath95-boson sectors
@xmath158 are adiabatic invariants of the dynamics .
an initial state in @xmath158 of the form @xmath159 evolves into @xmath160 applying @xmath141 and using @xmath161 in the case of a neutral system yields ( [ adiapprox ] ) .
[ thm : foae ] for any @xmath147 there are constants @xmath148 and @xmath149 such that for all @xmath150 , @xmath151 and @xmath152 @xmath162 & & & & & & & & & & & & & & & & & & \leq c\,{\varepsilon}^2\ , \ln ( { \varepsilon}^{-1 } ) \,\sqrt{m+1 } \end{aligned}\ ] ] in @xmath154 and @xmath155 with @xmath163 and @xmath164 note the infrared regularization by the characteristic function @xmath165 cutting off boson momenta smaller than @xmath166 in this definition , which can be omitted for neutral systems . with the help of this theorem
we can define a decomposition of the wave function into a superadiabatic part @xmath167 and the non - adiabatic part @xmath168 with different photon number . for a wavefunction @xmath169
we define @xmath170 and @xmath171 our final result concerns the amount of energy lost by the system due to radiation . for the stationary problem with hamiltonian @xmath54 independent of time , the natural definition for the energy carried by free photons in the state @xmath172 is @xmath173 where @xmath75 is the ground state energy . for
time - dependent hamiltonians the definition of energy is somewhat subtle , more so , because we look at the energy of a subsystem , the free photons .
however , in the superadiabatic representation there is again a natural definition for the energy given by the free photons , namely @xmath174 we will show that at any time @xmath9 when @xmath175 , it holds that @xmath176 thus in a situation where the sources move only during a finite time interval , before and after the change the definition of energy agrees with the static one .
note that the error of order @xmath177 comes from the infrared regularization in the transformation @xmath141 and is identically zero for neutral systems .
the equality ( [ energy ] ) is also determining the exponent @xmath178 in the infrared regularization .
a smaller exponent there would lead to a larger error in ( [ energy ] ) .
if we assume that the initial state is the dressed vacuum , then the energy emitted as a result of the motion of the sources is just the energy of the free photons and thus @xmath179 , which is computed explicitly in the following theorem .
[ thm : rad ] let @xmath180 , then uniformly on bounded intervals in time it holds that @xmath181 where @xmath182 is the second derivative of the dipole moment .
we call @xmath183 the subspace of the fock space for which @xmath184 for all but finitely many @xmath95
. the second quantization of a self - adjoint multiplication operator @xmath185 with domain @xmath186 is defined for @xmath187 with components @xmath188 as @xmath189 and is essentially self - adjoint . in particular , the free field hamiltonian @xmath190 is self - adjoint on its maximal domain .
the annihilation operator and the creation operator on @xmath187 are defined by @xmath191 here @xmath192 means that @xmath193 is omitted .
they fulfill the canonical commutation relations @xmath194=\langle f , g \rangle_{l^2({{\mathbb{r}}}^3)}\ , , \ \ [ a(f),a(g)]=0\ , , \ \ [ a^\dagger(f),a^\dagger(g)]=0\,.\ ] ] the creation and annihilation operator define the segal field operator @xmath195 it is essentially self - adjoint on @xmath196 .
the canonical commutation relations imply @xmath197&={\mathrm{i}}\ , \mathrm{im } \langle f , g \rangle_{l^2({{\mathbb{r}}}^3 ) } \quad\mbox{and}\quad [ { { \mathrm{d}}}\gamma(\omega),{\mathrm{i}}\phi(f)]=\phi({\mathrm{i}}\omega f)\,.\end{aligned}\ ] ] as described in the introduction we consider the hamiltonian @xmath198 with @xmath199 defined in ( [ equ : v(x(t),k ) ] ) .
as to be stated more precisely in lemma [ lem : spc ] and after , this hamiltonian has a ground state only for a neutral system , i.e. if @xmath200 , otherwise the bottom of the spectrum is not an eigenvalue . to deal with this fact we introduce an infrared cutoff @xmath201 in the interaction and put @xmath202 and @xmath203 .
the resulting truncated hamiltonian @xmath204 is a good approximation to the original hamiltonian as will be proven in proposition [ prop : approxs ] .
we emphasize that @xmath205 is used only as a tool in the proofs and the final results of section 2 hold for the hamiltonian @xmath206 without infrared cutoff .
there exists a standard estimate for @xmath46 ( cf .
e.g. @xcite proposition 1.3.8 ) .
let @xmath212 @xmath213 and @xmath214 , then @xmath215 thus @xmath216 is infinitesimally bounded with respect to @xmath210 if @xmath217 equation holds uniformly for @xmath218 and @xmath219 $ ] .
hence @xmath205 is self - adjoint on @xmath220 by the kato - rellich theorem ( cf .
@xcite , theorem x.12 ) for all @xmath219 $ ] and the graph norms are all equivalent .
1 . @xmath223 2 .
@xmath224 generates a strongly continuous unitary evolution family denoted by @xmath225 .
moreover , @xmath225 is a bounded operator on @xmath226 with @xmath227 } { \left\|u_{h^{\sigma}}(s , t_0)\right\| } _ { \mathcal{l}(d(h_{\rm f } ) ) } < \infty\ ] ] for any @xmath78 and uniformly in @xmath211 .
according to we need to show that @xmath228 which obviously holds under our assumptions . that @xmath229 and @xmath224 generate unitary evolution families follows from the first statement and the general result about contraction semigroups ( e.g. @xcite , theorem x.70 ) .
the norm bound @xmath230 in @xmath231 is less known and was proved in theorem 5.1.(e ) of @xcite .
[ lem : spc ] let @xmath232 with @xmath233 such that @xmath234 then the spectrum of @xmath235 is given by @xmath236 and the infimum of the spectrum @xmath237 is an eigenvalue if and only if @xmath238 in this case the unitary operator @xmath239 is well - defined and @xmath240 . in this section
we introduce the perturbed projections and the transformation which diagonalizes the infrared regularized hamiltonian up to @xmath248 . as explained in the introduction
the idea is to modify the adiabatic projections @xmath249 on the @xmath95-free - photons subspaces in such a way , that the modified superadiabatic projections @xmath250 are higher order adiabatic invariants , i.e. that @xmath251=\mathcal{o}({\varepsilon}^{2})\]]holds .
this construction is by now standard , see @xcite , and yields @xmath252 v_\sigma(t ) + { { \mathcal{o}}}({\varepsilon}^2)\,,\ ] ] where we introduce the shorthands @xmath253 = \phi\big({\mathrm{i}}\sum_{j=1}^n \frac { e_j \hat \varphi_\sigma(k)}{|k|^{{3 /2}}}\;{\mathrm{e}}^ { { \mathrm{i}}k \cdot x_j(t ) } \langle \kappa,\dot x_j(t)\rangle\big)=:\phi(z_2(t))\,.\end{aligned}\ ] ] here @xmath254 is the unit vector in the direction of @xmath255 .
the canonical commutation relations link @xmath256 and @xmath257 in the following way @xmath258=\phi_1\,.\ ] ] now the corresponding dressing operator @xmath259 that maps @xmath260 to @xmath94 then needs to have the expansion @xmath261 this suggests to define @xmath262 [ lem : v ] for @xmath263 $ ] and @xmath264 $ ] the operator @xmath103 is unitary , belongs to @xmath265 and satisfies @xmath266 uniformly in @xmath211 and @xmath7 . + the map @xmath267
, @xmath268 is differentiable and @xmath269-\tfrac{{\varepsilon}^2}{2 } [ \phi_2,\dot\phi_2 ] \right)\ , v^{\varepsilon}_\sigma \,.\ ] ] with @xmath270\;=\;v_\sigma(t ) \left ( v_\sigma^*(t ) h_{\rm f } v_\sigma(t ) - h_{\rm f}\right ) \;=\ ; v_\sigma(t ) ( \phi_0(t)-e_\sigma(t))\ ] ] and @xmath271&=&{{\mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2 } \left ( { { \mathrm{e}}}^{-{\mathrm{i}}{\varepsilon}\phi_2 } h_{\rm f } { { \mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2}- h_{\rm f}\right ) \\&= & { { \mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2 } \big ( \sum_{j=0}^\infty { \varepsilon}^j\,\frac{[-{\mathrm{i}}\phi_2 , [ -{\mathrm{i}}\phi_2 , [ -{\mathrm{i}}\phi_2,\cdots h_{\rm f}]]]]}{j!}-h_{\rm f } \big ) \\ & = & { { \mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2}\left ( { \varepsilon}\phi_1 - { \mathrm{i}}\tfrac{{\varepsilon}^2}{2 } { [ \phi_2 , \phi_1 ] } \right)\end{aligned}\ ] ] the first statement follows , since @xmath272 , @xmath256 and @xmath273= { \mathrm{i}}\,{\rm im}\langle z_2 , z_1 \rangle$ ] are bounded independently of @xmath211 as operators from @xmath36 to @xmath274 . under our hypotheses on @xmath45 the operators @xmath275 and @xmath276 commute as a consequence of ( [ ccr1 ] ) and @xmath277 .
hence @xmath278 for the other exponential we find @xmath279 \;=\ ; \left ( \partial_t - { { \mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2}\partial_t{{\mathrm{e}}}^{-{\mathrm{i}}{\varepsilon}\phi_2}\right ) { { \mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2 } \;=\ ; \left ( { \mathrm{i}}{\varepsilon}\dot\phi_2 -\tfrac{{\varepsilon}^2}{2 } [ \phi_2,\dot\phi_2 ] \right){{\mathrm{e}}}^{{\mathrm{i}}{\varepsilon}\phi_2}\ , .
\end{aligned}\ ] ] finally @xmath280\ ] ] implies ( [ abl ] ) . with the help of the dressing transformation @xmath103
, we define the dressed hamiltonians @xmath281 { \mathrm{i}}{\varepsilon}\partial_t -h^\sigma_{\mathrm{dress}}(t)&:=&v^{\varepsilon}_\sigma ( t ) \big({\mathrm{i}}{\varepsilon}\partial_t -h^{\sigma}(t ) \big)v^{\varepsilon}_\sigma(t)^*\ , .
\end{array}\ ] ] as a consequence of lemma [ lem : v ] the dressed hamiltonians are self - adjoint on @xmath213 since @xmath103 is a bijection on @xmath213 .
the corresponding evolution families are related through @xmath282 for @xmath285 we find @xmath286]]}{j!}+ e_\sigma \\ & = & h_{\rm f } + e_\sigma - { \varepsilon}\phi_1 - { \mathrm{i}}\tfrac{{\varepsilon}^2}{2 } { [ \phi_2 , \phi_1 ] } \;\;=\;\ ; h_{\rm f } + e_\sigma - { \varepsilon}\phi_1 + \tfrac{{\varepsilon}^2}{2 } { \rm i m } \langle z_2,z_1\rangle\ , , \end{aligned}\ ] ] and with ( [ abl ] ) @xmath287+\tfrac{{\mathrm{i}}{\varepsilon}^3}{2 } [ \phi_2,\dot\phi_2]\,.\ ] ] hence @xmath288 by definition we have @xmath292 let @xmath293 then bosons with momenta smaller than @xmath211 evolve freely under the regularized evolution , i.e. @xmath294 holds for all @xmath78 . to see this note that the equality holds for @xmath295 and the derivative of the left hand side vanishes since @xmath296 = [ h_{\rm f } , \phi({{\mathrm{e}}}^ { { \mathrm{i}}|k| \frac{t}{{\varepsilon } } } f_{\leq\sigma}(s ) ) ] = -{\mathrm{i}}{\varepsilon}\tfrac{{{\mathrm{d}}}}{{{\mathrm{d}}}t } \phi({{\mathrm{e}}}^ { { \mathrm{i}}|k| \frac{t}{{\varepsilon } } } f_{\leq\sigma}(s))\,.\ ] ] moreover , @xmath297 commutes with @xmath298 . by the duhamel formula we thus find @xmath299 the following lemma , which will be applied several times in the following , together with the fact that @xmath300 uniformly for @xmath301 $ ] allows us to conclude . in this section
we first show that the statements of theorems [ thm : adapp ] and [ thm : foae ] hold for the infrared regularized evolution @xmath307 with an error depending only logarithmically on @xmath211 .
then we use proposition [ prop : approxs ] and an appropriate choice for @xmath308 to show the statements also for the full evolution @xmath309 . clearly
the statement of theorem [ thm : adapp ] is a consequence of theorem [ thm : foae ] , but since the proof necessarily proceeds in the same two steps , we separated the statements . using again the duhamel argument together with proposition [ prop13 ] we find @xmath323 in the norm of @xmath324 . inspecting ( [ v2dot ] )
shows that @xmath325 hence ( [ uu2 ] ) follows , once we show that the second term does not contribute to the leading order transitions .
this follows from an integration by parts , @xmath326 now theorem 1 and 2 follow from proposition [ lem : letzte ] and proposition [ prop : approxs ] . to see this put @xmath327 and define @xmath328 then @xmath329 & =
\ ; { { \mathcal{o}}}({\varepsilon}^7\,\sqrt{m+1 } ) + { { \mathcal{o}}}({\varepsilon}^2 \ln({\varepsilon}^{-1})\,\sqrt{m+1})\,.\end{aligned}\ ] ] the probability for emitting a photon until time @xmath9 is thus given by @xmath337 .
but this has no simple asymptotics for @xmath121 because of the infrared problem .
however , for the radiated energy there is a simple asymptotic expression .
recall the definition of the energy of free photons in .
note first that at any time @xmath9 when @xmath175 , it holds that @xmath338 and @xmath339 and thus with corollary [ cor : spc ] @xmath340 in the third equality we used equation to bound @xmath341 by @xmath342 and ( [ esigma ] ) .
this shows that whenever the energy of free photons is unambiguously defined , our definition using the superadiabatic approximation agrees with it . we will now prove theorem [ thm : rad ] by plugging the superadiabatic approximation into the definition . note
that this computation of the energy is the reason that we insisted on all our estimates being valid also in @xmath343 . for observe that with the splitting given in corollary [ thm : cta ] we have that @xmath344 where @xmath345 just stands for the explicit expression claimed in . for the last term @xmath346 follows from @xmath347 and for the mixed term @xmath348 from integrating @xmath349 by parts as in the proof of proposition [ prop : approxs ] . in order to get also , we first transform the integral with @xmath350 , @xmath351 and define @xmath352 .
the resulting integral is @xmath353 where we abbreviated @xmath354 , @xmath355 and @xmath356 .
note that @xmath357 and @xmath358 are uniformly bounded on the domain of integration .
thus we can replace @xmath359 by @xmath360 with an error of order @xmath361 , which is negligible . observing that after summation the integrand is symmetric with respect to change of sign in @xmath362 and @xmath363 ,
i.e. for @xmath364 it holds that @xmath365 , we can extend the @xmath362-integration to all of @xmath366 at a cost of a factor @xmath367 .
modulo higher order terms this leads to @xmath368 where @xmath369 denotes the fourier transformation .
since with @xmath40 also @xmath370 is a schwartz function and since @xmath357 and @xmath358 are uniformly bounded on the domain of integration , we can shrink @xmath371 to @xmath372 with an error that is asymptotically smaller then any inverse power of @xmath372 in the region where @xmath373 .
the same replacement in the region where @xmath374 leads to an error of order @xmath375 . taking
e.g. @xmath376 leads to negligible errors .
now we can taylor expand the integrand as @xmath377 and the remaining leading order contribution is @xmath378 with @xmath379 . with the same argument as above we can replace @xmath380 by @xmath381 while making an negligible error and end up with @xmath382 v. betz and s. teufel .
landau - zener formulae from adiabatic transition histories . in _ mathematical physics of quantum mechanics _ ,
volume 690 of _ lecture notes in physics _
, pages 1932 .
springer berlin / heidelberg , 2006 . | we consider the dynamics of a massless scalar field with time - dependent sources in the adiabatic limit .
this is an example of an adiabatic problem without spectral gap .
the main goal of our paper is to illustrate the difference between the error of the adiabatic approximation and the concept of non - adiabatic transitions for gapless systems . in our example
the non - adiabatic transitions correspond to emission of free bosons , while the error of the adiabatic approximation is dominated by a velocity - dependent deformation of the ground state of the field . in order to capture these concepts
precisely , we show how to construct super - adiabatic approximations for a gapless system . |
protein - protein interactions play key roles in many cellular processes such as signal transduction , bioenergetics , and the immune response @xcite . moreover , many proteins function in the context of protein complexes of variable sizes and lifetimes
. examples of such complexes are ribosomes , polymerases , spliceosomes , nuclear pore complexes , cytoskeletal structures like the mitotic spindle or actin stress fibers , adhesion contacts , the anaphase - promoting complex , and the endocytotic complex @xcite . for yeast , 800 different core complexes
have been identified , suggesting the existence of 3000 core complexes for humans @xcite .
in addition it has been shown for yeast that most protein complexes are assembled just - in - time during the course of the cell cycle @xcite .
in fact many protein complexes in the cell are highly dynamic , with fast turnover of many components .
one can argue that their dynamics , although experimentally very hard to access , is biologically more relevant than their equilibrium properties .
therefore a systematic understanding of the dynamics of protein complexes in cells is one of the grand challenges in quantitative biology .
the elementary unit of all of these cellular processes is the bimolecular protein - protein interaction .
the strength and specificity of protein - protein association are determined by the integrated effect of different interactions , including shape complementarity , van der waals interactions , hydrogen bonding , electrostatic interactions and hydrophobic effects .
for example , the importance of electrostatic interactions has been demonstrated by experimental measurement at different ionic strengths @xcite . to a first approximation ,
bimolecular reactions are characterized by on- and off - rates .
the equilibrium association constant ( or affinity ) then follows as the ratio of the two . from a conceptual point of view , on- and off - rates are very different . on - rates
are commonly believed to be controlled by the diffusion properties as well as by long - ranged electrostatic interactions , whereas off - rates are rather controlled by short - ranged interactions like hydrogen bonding and van der waals forces .
the main features of the dynamics of protein association can be conceptualized within the framework of the _ encounter complex _ @xcite . to this end
, the association is divided into two parts .
first , mutual entanglement - the encounter complex - is achieved by the proteins due to a transport process including mainly diffusion but also electrostatic steering on small length scales @xcite .
if diffusion - controlled , classical continuum approaches can be used to describe this part of the process @xcite . to form the final complex , the system then has to overcome a free energy barrier due to local effects like dehydration of the binding interface @xcite . due to the various molecular contributions involved in this step , here the two binding partners essentially have to be modelled at atomic detail .
moreover the solvent may need to be treated explicitly and one might has to account for conformational changes @xcite .
thus , it appears reasonable to use the encounter complex as a crossover point from a detailed , atomistic treatment to a coarse - grained model and vice versa .
thermal fluctuations are an essential element of protein - protein encounter because they allow the two partners to exhaustively search space for access to the binding interface . from the viewpoint of stochastic dynamics ,
protein - protein association is a first passage time problem which can be addressed mathematically in the framework of langevin equations .
the application of langevin equations to association phenomena goes back to early work in the colloidal sciences @xcite . in these early approaches ,
the reactants were considered to have small spatial extensions and to be uniformly reactive . for large biomolecules like proteins ,
the situation is fundamentally different .
typically , proteins and other biomolecules have specific sites on their surface , where a particular binding reaction can take place . therefore , such binding events are subject to intrinsic geometric constraints for every particular protein - protein pair or larger assembly . the standard model for ligand - receptor interaction
was introduced by berg and purcell in the context of chemoreception based on the idea of using reactive patches to model anisotropic reactivity @xcite . due to anisotropic reactivity ,
also rotational diffusion becomes important .
shoup et al . showed that the effect of rotational diffusion can strongly increase the association rate between a receptor with a flat reactive patch and uniformly reactive ligands @xcite .
later , analytic expressions for the association rate between two spherical particles with both carrying a flat axially symmetric @xcite and asymmetric @xcite reactive patch were derived .
similar concepts were also applied by schulten and colleagues @xcite .
for many important aspects , analytical approaches are not possible and computer simulations are required .
this approach has been used early for protein - protein association @xcite .
the importance of electrostatic interactions for long ranged attraction was also emphasized by brownian dynamics simulations of protein - protein encounter @xcite .
if atomic structure is taken into account , then successful encounters are defined by simultaneous fulfillment of two to three distance conditions between opposing residues on the two surfaces @xcite .
brownian dynamics have also been used for the simulation of high density solutions , e.g. by bicout and field who studied a cellular `` soup '' containing ribosomes , proteins and trna molecules @xcite , or by elcock and coworkers who simulated a crowded cytosol for 10@xmath0s @xcite . in order to develop a quantitative framework for modelling the dynamics of protein complexes
, it is essential to understand the relative importance of generic principles and molecularly determined features of specific systems of interest .
only a good understanding of this issues will allow us in the future to develop reasonable coarse - graining strategies to address also large complexes of biological relevance . in this study
, we therefore address how general principles guiding the diffusional association of biomolecular pairs are modulated by their particular physicochemical properties . to this end
we have selected three molecular systems of interest with different steric and electrostatic properties .
one of the best studied bimolecular complexes is the extracellular ribonuclease barnase and its intracellular inhibitor barstar .
both proteins carry a net charge of @xmath1 and @xmath2 , respectively , which leads to a considerable electrostatic steering @xcite . considering the structure of the two proteins , barnase has a bean - like form , matching well on a large reactive area with the nearly spherical barstar .
a classic example of electrostatically - driven protein association is the iso-1-cytochrome _ c _ - cytochrome c peroxidase ( cytc : ccp ) complex , charged with @xmath3 and @xmath4 , respectively , and exhibiting dipoles aligned well with the reactive areas @xcite .
finally , we selected the medically important complex of a peptide fragment of p53 and its inhibitor mdm2 , which is used for anticancer drug design . in this system , electrostatic attraction plays a minor role . on the other hand ,
the steric match of the two surfaces is of particular importance here .
it is a perfect example of a key - lock binding interface , where p53 is buried deep into a cleft on the mdm2 surface . in this paper
we systematically explore the effect of various coarse - graining procedures on the rate for protein - protein encounter for the three selected model systems .
we revisit early approaches based on langevin equations and combine them with current knowledge on molecular structure .
the paper is organized as follows : in sect.[model_and_methods ] , we present our different stochastic models and describe the methods we use to parameterize the three considered bimolecular model systems . sect . [ results ]
contains the main findings of our study , which are discussed and summarized in sect.[summary_and_discussion ] .
one aim of this work is to determine how important specific details of the model proteins are with respect to the association properties .
therefore , we considered three different levels of detail as depicted in fig .
[ different_model_scheme ] for the three chosen systems . in the most generic approach ( @xmath5 ) , we only considered the steric interaction between spherical particles covered with reaction patches .
as a first refinement ( @xmath6 ) , an effective coulombic interaction was introduced using the dipolar sphere model ( dsm ) .
finally , since our langevin equation approach is particularly suited to capture anisotropic transport , we consider a more refined version for protein sterics ( @xmath7 ) . in this approach
the excluded volume of each protein was modeled by 8 - 25 smaller beads .
@xmath7 uses the dsm as well . in fig .
[ different_model_scheme ] , we also show the full structures in the bottom row as surface representations , including the locations of the binding interfaces . in the following , the general properties of the simulation model and the different techniques used in this work will be explained .
the diffusion of the protein model particles is described by an anisotropic @xmath8 diffusion matrix in all versions of our model . in ref .
@xcite , de la torre and coworkers present a method to calculate this diffusion matrix from the pdb structure of a protein .
this method has been implemented in a software called hydropro which is provided online by the same authors ( http://leonardo.fcu.um.es/macromol ) .
the basic concept is to put spheres of a certain size at the position of any non - h atom .
the volume of these spheres effectively models a fixed hydration shell .
this construct is then filled up with smaller , densely packed , but non - overlapping spheres .
since the hydrodynamic properties of a rigid body are determined by its outer boundary only , a shell of these small spheres is generated by deleting all spheres which have a maximum number of possible neighbors . to this shell a sophisticated technique is applied , which has been developed by de la torre and colleagues over the years , to calculate the diffusion matrix of such a cluster of non - overlapping spheres ( see references in @xcite ) .
several system properties are implicitely contained in the mobility matrix , such as ambient temperature @xmath9k , as well as the density and dynamic viscosity of the solvent , where we chose the respective parameters of water , @xmath10g@xmath11@xmath12 and @xmath13pa s. for simplicity , hydrodynamic interactions were not introduced in our models , because the corresponding effect on the association rates is expected to be well below 10% @xcite . for the integration of the langevin equation , which describes the stochastic motion of the particles , we follow an approach which has been recently developed to model cell adhesion via reactive receptor patches @xcite .
let @xmath14 be a six - dimensional vector describing position and orientation of a particle at time @xmath15 .
since the noise due to brownian motion is additive ( which means that it does not depend on @xmath14 due to a constant mobility matrix @xmath16 ) , the langevin equation is given by : @xmath17 here , @xmath18 is a six - dimensional vector containing the force and torque acting on the particle , and @xmath19 denotes gaussian white noise : @xmath20 as explained in app .
c of ref .
@xcite , the euler algorithm can be used to solve a discretized version of this equation : @xmath21 for proteins , the typical orders of magnitude are @xmath22@xmath23s@xmath24 and @xmath25 nm .
therefore , a reasonable choice for the time step is @xmath26ps , as this leads to a mean step length of @xmath27 nm .
the mobility matrix of a particle is defined in a particle - fixed coordinate system .
thus , the whole step has to be calculated in terms of particle - fixed coordinates and then transformed to the laboratory coordinate space . in particular
, this transformation implies a rotation @xmath28 regarding the orientation of the particle .
special attention has to be payed to the force @xmath18 , which is typically calculated in the global frame of reference and hence has to be transformed to particle space before eq .
[ langevin_equation_discrete ] can be evaluated .
this back - transformation is achieved by applying @xmath29 to @xmath18 .
since rotation matrices @xmath28 simply consist of a list of orthonormal vectors , their inverse is equal to the transposed matrix @xmath30 .
thus eq.[langevin_equation_discrete ] can be rewritten as @xmath31 + { \mathcal o}(\delta t^2)\text { . }
\label{langevin_equation_discrete_rotated}\ ] ] as @xmath18 and @xmath32 are six - dimensional and contain information about torque and rotation , eq .
[ langevin_equation_discrete_rotated ] is only formally correct , as @xmath28 acts on both the translational and rotational parts of the respective vectors separately . in each step of the simulation , a displacement vector @xmath33
is drawn for each particle as described above .
if this global displacement leads to any violation of the hardcore repulsion , all suggested displacements are rejected and new @xmath34 are calculated .
this procedure continues , until an update of all positions and orientations is found which does not lead to any overlap . in this way , the constraint according to the excluded volume effect is included in the stochastic motion .
the spherical reactive patches are not taken into account for the steric interactions , i.e.they may not only overlap pairwise but also with the model particles .
one would expect that our procedure leads to errors of order @xmath35 if two particles are in close proximity of order @xmath36 .
however , it has been shown for a different system @xcite that in practise the deviation from the expected behavior is very small and thus the approach is reasonable . as mentioned before
, the @xmath8 mobility matrix @xmath16 represents anisotropic diffusion . for large times ,
anisotropic diffusion crosses over into isotropic diffusion because the information about the initial orientation gets lost after a certain relaxation time due to the rotational diffusion @xcite .
in general , translational and rotational diffusion are coupled so that large time steps can not be used . however , for the particular systems studied here , we found that the diffusive coupling is a very small effect .
in particular , the major entries in the diffusion matrix of the proteins used here according to hydropro multiplied with different powers of the stokes radius @xmath37 cm to make the dimensions comparable are @xmath38 , @xmath39 , @xmath40 .
therefore , the effect of diffusive coupling is @xmath41 and @xmath42 smaller than rotational and translational diffusion , respectively . finally , the typical time scale at which the cross - over is expected can be calculated to be @xmath43ns .
time steps of this magnitude were rarely used in the simulations ( see below ) , so that for most of the steps , the anisotropicity is well preserved .
therefore we can safely neglect changes in the anisotropicity of the mobility matrix .
the simulations were performed in a cubic box with periodic boundary conditions .
schreiber and fersht used concentrations between @xmath44 m and @xmath45 m in their experimental studies of the association rate of the barnase : barstar complex @xcite . the average volume containing one particle at a concentration @xmath46 is @xmath47 with the avogadro number @xmath48mol@xmath24 .
hence , the edge length of a cubic boundary box representing concentration @xmath46 can be calculated from @xmath49{v}=1/\sqrt[3]{c n_a}$ ] .
e.g. @xmath50 m leads to @xmath51 for one pair of particles , which is two orders of magnitude larger than the size of the proteins . due to this low density
, the first passage times ( fpt ) for encounter can be expected to be much longer than the chosen time step . for computational efficiency
, we therefore used a variable time step in our simulations .
van zon and ten wolde suggested a method to avoid unwanted collisions when they introduced their green s function reaction dynamics ( gfrd ) @xcite .
in contrast to our work , however , this method is based on isotropic diffusion .
generalizing the gfrd to anisotropic diffusion is out of the scope of our work and we therefore used the following scheme .
we first note that in gfrd each time step is chosen such that it includes the next reaction . in our case , we also want to investigate the stochastic dynamics before the next encounter event takes place .
thus a large time step is not chosen to include the next encounter , but to bring the system to such a configuration that encounter becomes more likely .
this step can be well represented by isotropic diffusion with an overall diffusion constant @xmath52 following from the anisotropic diffusion matrix . for an isotropic random walk , the displacement probability is given by a gaussian distribution with spherical symmetry .
thus , large spatial steps are exponentially suppressed , which makes a step of size @xmath53 an @xmath54-sigma event . by setting @xmath55 the smallest effective particle distance in the system , where effective means the distance of the surfaces @xmath56 with @xmath57 determining the maximal steric interaction radius of particle @xmath58 , one can estimate a reasonable time step for which a collision is highly improbable .
van zon and ten wolde found that the choice @xmath59 provides good results , combined with the fact that wrongly sampling a collision event would need a certain direction of the displacement in addition to the length .
as the particles reach close proximity @xmath60 , the estimated @xmath35 vanishes and thus the simulation would be slowed down infinitely .
therefore , there has to be some lower boundary for the time step @xmath61 , which is generally chosen to be @xmath62ps as explained earlier .
thus , the adapted time step is given by : @xmath63 in practice , most time steps are in the ps - range , with very few time steps coming up to the ns - range .
electrostatic interactions are known to play an important role in protein association . to study the effect of electrostatics in our generic model , the models @xmath6 and
@xmath7 utilized the dipolar sphere model ( dsm ) , following refs .
the dsm effectively models a monopole and dipole interaction by summing over the interactions of three charges , one positioned in the center of each particle , and two close to its surface in opposite positions .
taking into account the debye screening function due to the presence of counter ions in solution , the electrostatic interaction energy between two charges @xmath64 at positions @xmath65 respectively with distance @xmath66 is : @xmath67 here , @xmath68 is the inverse debye screening length , which typically has a value of @xmath69 nm under physiological conditions .
we assume a value of @xmath70 for the relative static permittivity of the medium , which reflects the properties of water at ambient temperature .
@xmath71 is a correction to the screening of charges which are placed in an object like a protein which has no free charges inside .
taking @xmath72 as the closest distance of @xmath64 from the surface of the surrounding protein , it is approximately given by @xmath73 .
this potential leads to a force of charge @xmath74 on @xmath75 : @xmath76 as our simulation uses periodic boundary conditions , actually an infinite number of copies exists for every charge .
however , due to the very fast decay of the screened electrostatic interaction , only the minimum image distance of two charges is considered in the force calculation .
two model particles @xmath77 and @xmath78 feel the sum of the coulomb forces @xmath79 between all pairs of the three complementary charges mimicking the monopolar and dipolar interactions .
thus the full force between particle @xmath77 and @xmath78 is @xmath80 , where @xmath58/@xmath81 run over the charges of @xmath77/@xmath78 respectively .
as explained earlier , the action of the force on a particle in the langevin equation is weighted with the mobility matrix @xmath16 .
the hydropro software directly gives the diffusion matrix @xmath82 .
this means that in our case the force action should be rewritten as @xmath83=@xmath84 .
considering a time step of @xmath85ps and a typical diffusion constant @xmath86 , we have @xmath87 m and @xmath88 m for typical distances
@xmath89 nm and @xmath90 nm , respectively .
in contrast , the typical step length due to the brownian motion is @xmath91 m .
this shows that the magnitude of electrostatic interactions at distances of @xmath92 nm is much smaller than thermal energy .
therefore the effect of force is also not considered in our adaptive scheme for the time steps .
however , it can be expected that the _
systematic _ drift , albeit small , will still lead to an altered encounter behavior .
gabdoulline and wade @xcite used several criteria to define contact areas of bimolecular protein complexes . in our studies , we define the contact area to consist of those atoms in the two interacting proteins that are at @xmath93 or less distance from an atom of the complementary protein . the center of mass of these atoms is considered as the center of the reactive area . for @xmath5 and @xmath6
, the reactive patch is centered at the surface of the sphere modeling the excluded volume such that it has the same relative direction from the center of mass as obtained by the method described . in the case of @xmath7 , the center of the patch
is set to the center of the reactive area .
the contact area has a diameter of approximately 10 to 20 for the three systems studied here . following earlier brownian dynamics simulations with atomistic details @xcite we have performed an in - depth analysis of the free energy landscape and the encounter state of the protein complexes considered in this work ( unpublished results ) .
this showed that the encounter complex is typically located at relative separations of the two protein surfaces of about 10compared to their positions in the final complex .
as the spherical reactive patches used in this study simultaneously determine both the size of the contact area on the surface and the distance above their surface at which an encounter will be possible , values in the range of 5 to 10 seem to be reasonable .
note that as long as physical considerations do not dictate non - spherical reaction patches , the spherical choice is highly favorable for computational efficiency .
as already stated in the beginning , two types of excluded volume structures are taken into account . in the first case , used in @xmath5 and @xmath6 ,
the proteins are assumed to have an approximately spherical form . the radius for the model spheres determining the hard core interaction follows as the radius of gyration of the protein , which is also calculated by the hydropro software .
the underlying data in the more detailed approach @xmath7 is obtained using the atob bead modeling software @xcite . in this way
, the three - dimensional structure of the proteins is modeled with a comparably small number of 8 to 25 spheres of different sizes .
the monopole charge is the sum of all elementary charges in a protein and is placed at the center of the respective model particle .
the dipole moment @xmath94 is obtained by summing over the product of all atomic charges due to the xyz force field and their relative position to the center of mass . in the model , it is represented by two opposing charges which are positioned along the direction of @xmath95 and at a distance @xmath96 . the magnitude @xmath97 is chosen such that @xmath98 .
we found that the particular choice of @xmath99 does not have a noticeable influence on the results .
the resulting parameterization is given in tab .
[ parameter_table ] for the proteins considered here .
langevin dynamics simulations were performed for cubic boxes containing two model proteins .
simulations were conducted until the encounter condition was met for the first time ( typically after milliseconds ) .
because in our langevin simulations we measure the distribution of first passage times ( fpt ) to encounter , from which we can deduce the mean first passage time ( mfpt ) @xmath100 , the encounter frequency is defined as @xmath101 .
this choice is motivated by the fact that for a poisson - like process , the distribution of first passage times is given by @xmath102 and therefore the encounter frequency @xmath103 indeed satisfies @xmath101 . as the preparation of a comparable experiment would never allow knowing the particular initial positions and orientations of the unbound proteins , it makes sense to average over the possible initial configurations in the computer simulations .
therefore , we started a large number of runs ( typically @xmath104 to @xmath105 ) with random initial positions and orientations for all involved model particles , under the constraint that the initial pairwise distance is at least large enough to prevent an immediate encounter .
the `` first passage '' is defined as the first overlap of two complementary reactive patches .
interestingly , due to this averaging the first passage process becomes poisson - like , see fig .
[ randinit_histogram ] .
the data show a clear exponential behavior .
this means that it is justified to use the notion of an `` encounter frequency '' , as the fpt distribution is indeed represented by a single stochastic rate .
the finite probability at small fpt is due to the possibility that the two model particles are already in close proximity when the simulation is started .
the large errors in the histogram at @xmath106 are caused by the fact that exponentially sized histogram bins were used to sample the behavior for small @xmath107 .
therefore , events hitting a particular bin are rare because of the small width of the bins at @xmath108 , which then leads to bad statistics in this domain . as the encounter process is purely diffusion limited in @xmath5 , one would expect the encounter frequency to scale linearly with concentration .
[ randinit_concentrations ] demonstrates for the barnase : barstar system that this is indeed the case .
hence , it is reasonable to always scale the encounter frequencies with the inverse concentration , as will be done for the rest of this work .
we will denote these rescaled quantities as _ encounter rate _ , i.e. the encounter rates have the dimension m@xmath24s@xmath24 . in summary , we have demonstrated here that our microscopic model leads to a well - defined macroscopic encounter rate . in
most of the simulations , only one instance of the final complex was considered , i.e. one model particle of each kind .
using such small systems could lead to undesired finite size effects .
we therefore considered the effect of having many particles in the simulation box .
[ randinit_finite_size ] shows the simulation results for the encounter frequency @xmath103 for an increasing number of barnase : barstar pairs , while keeping the size of the boundary box constant . in order to understand the expected effects ,
consider a system with molecules of barnase ( @xmath109 and @xmath110 ) and two molecules of barstar ( @xmath111 and @xmath112 ) randomly distributed over the boundary box .
the relative alignment of any pair of @xmath109s and @xmath113s is therefore random again . for a particular pair
the distribution of times to first encounter will thus look very similar to the case with a single pair in the box , which is a simple exponential decay with respect to the encounter frequency @xmath114 : @xmath115 $ ] .
the probability that , e.g. , the particular pair @xmath116 reaches encounter at a certain time @xmath15 before the three other possible pairs ( @xmath117 , @xmath118 , @xmath119 ) , is therefore : @xmath120 thus , the probability that any of the four possible particle pairs reaches encounter before the respective three other pairs do , is @xmath121 as just calculated , i.e. @xmath122 has again a poisson form like @xmath123 and @xmath124 . in general , for higher numbers of particle pairs @xmath125 , we expect to again find an exponential distribution of the time to first encounter with the encounter frequency @xmath126 .
this quadratic behavior is nicely confirmed by the data shown in fig.[randinit_finite_size ] , which suggests that even for small systems with only two particles , no severe finite size effects have to be expected . in particular , this rules out that larger numbers of particles lead to noticeable three - body interactions or hindering of the encounter process .
one feature of special interest which we can address with our langevin equation approach is the pathway through which the encounter is formed .
we dissected the encounter process into several parts as visualized in fig .
[ alignment_states_scheme ] . at the start of each run
, the systems were prepared in the unaligned state @xmath127 , as described earlier .
a state of close approach which however does not allow for binding is called @xmath128 .
the two model proteins will switch between states @xmath127 and @xmath128 a number of times @xmath125 , until they finally reach the encounter complex @xmath129 due to a favorable combination of translational and rotational diffusion . in the following ,
each occurance of @xmath128 will be termed a _
contact_. thus @xmath125 counts the number of unsuccessful contacts before the encounter is finally formed .
a separate set of simulations was performed to measure the distribution of @xmath125 .
furthermore we analyzed the distribution of return times @xmath130 .
this is the time it takes for two model proteins to get into contact again ( @xmath127@xmath131@xmath128 ) after having lost translational alignment ( @xmath128@xmath131@xmath127 ) , i.e. after they were in close proximity . finally , we determined the distribution of resting times @xmath132 in translational alignment @xmath128 before the two model particles separated again .
as an example , fig .
[ alig_no_example ] shows the distribution of @xmath125 for the barnase : barstar model system at @xmath133 m in the framework of @xmath5 .
surprisingly the distribution of the number of contacts has again a poisson form .
note that the number of unsuccessful trials in state @xmath128 can be rather large ( up to @xmath104 ) .
we also found that the distribution of @xmath125 is roughly independent of concentration .
this is reasonable , as after the two proteins were in contact once , the further encounter process is guided by returns to state @xmath128 and thus should be more or less independent of system size .
[ alig_onoff_example ] shows that the return time @xmath130 ( plotted with the plus - symbol ) is not exponentially distributed .
instead , it follows a power law @xmath134 and undergoes an exponential cutoff due to the finite size of the boundary box at large @xmath130 .
therefore , there is a high probability for very small return times , i.e.situations , where the two model proteins do not really separate , but immediately after loosing translational alignment ( @xmath128@xmath131@xmath127 ) get closer again ( @xmath127@xmath131@xmath128 ) .
the power law behavior of the return time is consistent with the problem of a random walk to an absorber in three dimensions @xcite . in principal , these two situations are equivalent since the relative motion of the two proteins while unaligned @xmath127can be approximately understood as an isotropic random walk , and the criterion for going over to translational alignment @xmath128 reflects an absorbing boundary in the configuration space of relative positions .
the distribution of resting times @xmath132 ( plotted with the cross - symbol in fig .
[ alig_onoff_example ] ) follows the same power law as @xmath135 , but the exponential cutoff occurs much earlier .
the reason is that here the cutoff is determined by the region in configuration space where the two model proteins are in state @xmath128 .
as this is much smaller than the whole volume of the boundary box , in which they are unaligned and therefore in state @xmath127 , a random walk in state @xmath128 will end earlier .
the differences we obtain in the distributions of @xmath132 and @xmath130 when using the variants @xmath6 and @xmath7 compared to @xmath5 are generally very small and unlikely to account for any deviations in the overall encounter rates .
also , the distribution of @xmath125 is always well described by a single exponential decay .
however , the inverse decay length @xmath136 significantly varies between the different situations .
therefore , changes in the overall encounter rate are mainly caused by a different probability for reaching state @xmath129 from state @xmath128 .
this is reasonable when considering that the interactions are strongly localized and can thus only act while the system is in the aligned state @xmath128 .
so far we have only considered barnase : barstar ( @xmath137 ) to demonstrate how our computational model works .
we now use our setup for a more comprehensive investigation .
in particular , we also apply our method to two other systems , cytochrome c and its peroxidase ( @xmath138 ) as well as the p53:mdm2 complex ( @xmath139 ) . those represent systems with different interface characteristics and where the role of electrostatics is either much stronger ( @xmath138 ) or much weaker ( @xmath139 ) than for @xmath137 . to this end
, all the previously described quantities were measured for 8 different concentrations @xmath140pm .
furthermore , to find out how the choice of the radius of the reactive patch affects the results , we used patch radii of @xmath141 and @xmath142 in addition to the initially considered value of @xmath143 . tab .
[ main_data_table ] lists the encounter rates @xmath103 as obtained from these simulations .
the rates are all roughly of the same order of magnitude . yet
several interesting qualitative features are readily apparent .
first , for decreasing patch sizes , the rates generally decrease .
second , this effect is weaker for @xmath6compared to @xmath5 , which basically means that the electrostatic attraction and orientation due to the dipole interaction are indeed enhancing the encounter .
the strongest effect of the electrostatic interaction is obtained for cytc : ccp , which is the system with the largest monopole and dipole and the best alignment of the directions of the dipoles and the reactive patches . on the other hand p53:mdm2
is nearly unaffected by the effective charges , due to its weak monopole charges and , additionally , an unfavorable alignment of the dipolar interaction and the reactive surface area .
furthermore , regarding the results with detailed steric structure @xmath7 , the effect on the rate is correlated with the deviations of the protein forms from the spherical excluded volume approach in @xmath5 and @xmath6 .
this deviation is smallest for cytc : ccp and largest for p53:mdm2 . the findings for the encounter rate @xmath103 are also reflected in the results for @xmath136 .
as expected , an increase in @xmath103 correlates with a decrease in @xmath136 .
the only exception is cytc : ccp observed in @xmath6 , which is also special in regard to the effect of patch size . here , the effective coulombic interaction is strongest and the dipole moment is best aligned with the reactive patches . therefore ,
having reached state @xmath128 once , the proteins do systematically orient towards @xmath129 , while they are additionally strongly steered back towards @xmath128 when loosing their translational alignment .
this behavior is the stronger the closer the model proteins have approached once
i.e. for the case of small patch sizes , where state @xmath128implies the smallest distance . while this only explains the inversion in the @xmath136 behavior as a function of patch size @xmath144 , @xmath103 is obviously still slightly decreasing with smaller patch sizes
. this can be explained by the fact that the time to the first approach of state @xmath128 is larger for smaller patches , as this implies a smaller relative distance .
this obviously compensates the fact that afterwards the encounter is formed even quicker , as reflected by the decreasing @xmath136 . the strong correlation between the encounter rate @xmath103 and the mean number of contacts @xmath136 is also evident from the correlation plot in fig.[correlation_plot ] .
indeed , @xmath145 seems valid for most of the different systems and models .
it is noteworthy that the prefactor is very similar in all cases .
basically , this means that one unsuccessful contact takes the same amount of time on average , no matter what the local details of the system are .
this gets more obvious recalling the distributions of the resting and return times @xmath132 and @xmath130 in fig .
[ alig_onoff_example ] , which shows that @xmath146 . as the average time for one contact will be approximately @xmath147 , it is dominated by @xmath130 , which is only marginally influenced by the local details of the system and the chosen model .
therefore it can be concluded , that for @xmath137 and @xmath139 the incorporation of a more detailed modeling approach influences @xmath103 and @xmath136 , but not the overall characteristics of the encounter process .
the only exceptions for the clear correlation of @xmath103 and @xmath136 are @xmath6 and @xmath7 for the case cytc : ccp ( @xmath138 ) , where @xmath103 is nearly independent of @xmath136 because of the strong electrostatic interaction .
this is consistent with the earlier finding , that the behavior of cytc : ccp is qualitatively different @xcite , as its electrostatic interactions would facilitate long - lived nonspecific encounters between the proteins that allowed the severe orientational criteria for reaction to be overcome by rotational diffusion .
for all three systems studied , in @xmath7 the smallest patch size @xmath142 leads to a somewhat artificial slowing down , because in this case an overlap of the patches is rather hindered by the beads modeling the protein structure .
we next address the dependence of the data on the size of the reaction patches in more detail .
this behavior is exemplary studied with the barnase : barstar model system . in fig .
[ patch_size_plot ] , the encounter frequency has been obtained from simulations for barnase : barstar - like model particles in the framework of @xmath5 at several concentrations @xmath148m@xmath149nm@xmath150nm@xmath149pm@xmath151 and varying patch sizes @xmath144 .
all values in the figure have been scaled with the concentration , which leads to data collapse .
it is obvious that as @xmath144 gets larger than @xmath152 at around @xmath153 , the reactive patch covers the whole model particle and we therefore cross over to the smoluchowski limit of isotropic reactivity , where @xmath154 .
however , at high densities and large @xmath144 , the patches span a large part of the simulation box of edge length @xmath155 , and do immediately encounter for a threshold value of @xmath156 , where the sum of the patch diameters @xmath157 equals the triagonal .
thus , the encounter frequency must diverge with @xmath158 , where we suppose @xmath159 , as the volume of configurational space without immediate encounter is decreasing with @xmath160 .
this assumption in addition with the smoluchowski behavior would lead to @xmath161 for large @xmath144 , which follows the data in fig .
[ patch_size_plot ] well ( black dashed lines ) . as already mentioned it is well known that the electrostatic interaction of proteins can severely increase the association rate . however , under physiological salt conditions , coulombic interactions are screened by counter ions in the solution on a small length scale of approximately 1 nm . thus , deviations from case @xmath5 without effective charges will only arise for small @xmath144 .
[ patch_size_electro_plot ] shows the results of respective simulations for @xmath6 compared to the results for @xmath5 , as considered before .
indeed , for large patch radii @xmath144 , the results are similar , while for smaller @xmath144 , the encounter rates in @xmath6 are clearly higher compared to @xmath5 .
however , the crossover to a power law behavior with roughly @xmath162 can be detected for very small @xmath144 , but at a prefactor of about 50 times larger than for @xmath5 .
the main goal of this work was to model protein encounter in a generic framework which allows us to include molecular details without making future upscaling to larger complexes impossible .
our model approach incorporates steric , electrostatic and thermal interactions of the proteins considered .
these interactions are thought to be the major factors governing protein encounter .
not included are conformational changes of the proteins upon association , related entropic terms , and the molecular nature of the surrounding solvent that becomes relevant at close distances .
the model parameters are extracted from the atomic structures available in the protein data bank by generally applicable protocols as described in sect.[model_and_methods ] . in principle
, these methods of data extraction can be fully automatized .
the biggest advantage of our coarse - grained model is the possibility to extend the simulations to large scales in terms of particle numbers , time and system size . in many of the earlier studies @xcite ,
the system was prepared already close to encounter and the overall association rate was then calculated via a sophisticated path - integral like procedure . in contrast , our simulations account for the whole process of diffusional encounter and is thus rather general , allowing for spanning large time scales via our adaptive time step algorithm . in particular ,
each set of simulations consists of @xmath104 to @xmath105 runs of lengths up to the order of seconds and could be performed on a standard cpu within hours of computer time .
being able to directly obtain the first passage times ( fpt ) of the encounter processes in our model allows to check the validity of several phenomenological assumptions .
first of all , the fpt distribution matched very well a poisson process with a single stochastic rate , as seen in fig .
[ randinit_histogram ] , which validates the notions of encounter and association .
moreover , our approach provides two ways of controlling the particle density and for both cases the results corresponded well to the expected scaling .
first , the concentration is inversely correlated with the size of the periodic boundary simulation box .
we show that the encounter rate grows linearly with the particle concentration .
furthermore , leaving the box size constant , the concentration can also be varied by adding a higher number @xmath125 of particles . considering only the _ first _ encounter of any of the possible complementary pairs of model particles , the mean first passage time to this event
is not only lowered by a factor @xmath125 but we show that the expected behavior is an enhancement of the encounter frequency by @xmath163 , which is nicely matched by the results of the simulations
. therefore we can conclude that the computational model studied here satisfies the general requirement of stochastic bimolecular association processes that describe binding by a single rate constant . to test our model against known results
we have chosen three well - known bimolecular systems with different characteristics .
the barnase : barstar complex is the gold standard for protein - protein association and characterized by relatively strong electrostatic steering .
the association of cytochrome c and its peroxidase is even more strongly affected by coulombic attraction . here , both proteins have a rather spherical form .
finally , the p53:mdm2 complex has a different characteristic with a very small net charge and a deep cleft perfectly matching the small peptide p53 , whose reactive area is therefore nearly spanning over its whole surface .
these model systems were purposely chosen to check whether our effective representations of the protein properties would lead to reasonable and significantly distinguishable results . indeed , this is the case as the discussion of the results in tab .
[ main_data_table ] in the respective section shows . when comparing the results for the encounter rates in tab.[main_data_table ] with previous studies from the field of bimolecular protein association , several aspects have to be kept in mind .
first , throughout this study , we do only consider the _
encounter _ of our model particles .
as explained in the beginning , the complete association of the complex still lacks the step over a final free energy barrier , which is due to effects such as the dehydration of the protein surfaces and thus requires more detailed modelling . in the framework of our approach , this final step could be modelled by a stochastic rate criterion , where the rate can be obtained by transition state theory from the energy landscapes characterized in atomistic calculations . in any case , any additional process to be included can only lower the values found in our study . in the work on barnase
: barstar by schreiber et al .
@xcite , the authors reported that the association between barnase and barstar is a diffusion - limited reaction .
the argument for this is that the association rates at high ionic concentrations in the solution , i.e. for the limit in which the electrostatic steering gets negligible , are clearly lowered by the addition of glycerol , which will lead to slower diffusion . assuming diffusion control ,
the reactive step over the final barrier should be kinetically unimportant , as generally discussed in ref .
indeed , we see that our results for the encounter rates lead to values in the correct order of magnitude of @xmath164m@xmath24s@xmath24 , which is similar to the experimental value obtained by schreiber et al . for the association constant of barnase : barstar @xmath165m@xmath24s@xmath24 at physiological salt @xcite .
however , the basal association rate , i.e. the rate at high ionic strength , is reported as @xmath166m@xmath24s@xmath24 from experiments @xcite . given that the association process of brn : brs is diffusion limited
, these findings should actually coincide with our values for @xmath5 .
but as we already discussed in the results section , in our simulations the influence of the effective electrostatics introduced in @xmath6 do not result in such a drastic change of the encounter behavior . in several earlier approaches ,
similar problems have been addressed by computational and analytical studies . in work by zhou and coworkers , basal encounter rates for particles with reactive patches
have been found to be @xmath167m@xmath24s@xmath24 @xcite and @xmath168m@xmath24s@xmath24 @xcite , that is closer to the basal rates reported by schreiber and coworkers .
it has to be noted that , in both cases , the patches were flat areas above the surface of the spherical model particles , which had a smaller angular extension compared to our cases , and especially required a much closer translational approach ( @xmath169 in @xcite ) to form the encounter .
if we expand the graph in fig.[patch_size_plot ] to smaller patch radii like @xmath170 , we also find basal rates in the order of @xmath168m@xmath24s@xmath24 .
also , the deviation between @xmath5 and @xmath6 , i.e. the influence of the effective electrostatics , is more prominent and could enhance the encounter rate by about two orders of magnitude , which is consistent with the findings in the previously cited work .
there , the effect of coulombic interaction is reflected with a boltzmann factor due to a pairwise coulomb energy .
this approach works well , as shown in ref .
@xcite , and has been recently used in a more complex model study of the energy landscape of protein - protein association @xcite .
any model for the reaction patches has to rely on results obtained from more detailed modeling .
the surface of a protein is typically densely covered by water molecules due to the hydrophilic nature of its surface .
this hydration shell has a thickness of about 3 and will therefore in principal hinder the approach of two proteins to distances below 6 . setting the encounter patches to values below this threshold of 3 would then mean that part of the dehydration would already have happened before the encounter is actually formed , which is probably hardly described by simple diffusion with drift . moreover , all of the considered protein systems feature distinct key - lock binding interfaces regarding the steric structure , apart from some flexibility due to intrinsic thermal motion
. therefore , it makes sense to represent the encounter area by a three - dimensional extended object rather than by a flat surface region .
indeed , the results of our studies show that our approach is capable of reproducing encounter rates in a reasonable order of magnitude , qualitatively reproducing generally expected features . in an in - depth investigation of the dependency of the encounter rate and the patch radius it
is shown , that the choice of the geometry of the reactive area is at least as crucial for the results as definition of the model interactions and its parameters . in principal
, one could think of the patch radius as a valuable tuning parameter to fit experimental results and the encounter kinetics in the computational model .
our approach makes it possible to observe general features of the encounter process .
in particular , we dissect the pathway to the encounter complex in several levels of alignment between our model proteins .
as we observe the full trajectory to encounter in our simulations , we are able to extract the number of unsuccessful contacts @xmath125 between the proteins until they finally reach a reasonably aligned state to bind .
the distribution of @xmath125 is again in all cases well described by a single exponential decay .
this behavior is not obvious as the probability of success for one contact is depending on several aspects of diffusion in a complex manner .
first , the closer the rotational alignment at the beginning of the contact is to the encounter state , the higher is the probability of success .
second , this initial alignment is also coupled to the last contact if the time in between @xmath130 is small .
finally , longer contact resting times @xmath132 also increase the probability of encounter .
it is interesting , that all these effects still lead to a simple poisson distribution of the number of contacts @xmath125 when averaging over the initial conditions as it is done in this work .
furthermore , we find that the distributions of these resting and return times can not be described by a poisson process , but are consistent with the expectations for a spatially constricted random walk in three dimensions .
we find that the particular mean fpt to encounter is in most of the cases directly proportional to the number of unsuccessful contacts .
this seems to be a very fundamental qualitative feature irrespective of the details of the proteins and the applied model .
however , for cytc : ccp the behavior is qualitatively different , which is consistent with earlier studies of this highly electrostatically steered complex . in summary , here
we have presented a langevin equation approach to protein - protein association which in principle allows us to combine long simulation times and large systems with molecular details of the involved proteins .
this first study has focused on bimolecular reactions and has proven that this approach is capable of reproducing known association rates with a reasonable dependance on the main parameters involved .
one special strength of our approach is that it allows us to address the details of the binding pathway , for example by measuring the statistics of unsuccessful contacts before encounter . in the future
, this approach will be extended to large protein complexes of special biological interest .
we thank christian korn for many helpful discussions .
this work was supported by the volkswagen foundation through grants i/80469 and i/80470 to v.h . and
, respectively .
j.s . and u.s.s
. are supported by the center for modelling and simulation in the biosciences ( bioms ) at heidelberg and by the karlsruhe institute of technology ( kit ) through its concept for the future .
.[parameter_table ] protein structures and parameters used in the study .
the coordinates of the patches ( @xmath171 ) and the dipole moment ( @xmath94 ) are given relative to the center of mass . [ cols="^,^,^,^,^,^,^,^",options="header " , ] | we study the formation of protein - protein encounter complexes with a langevin equation approach that considers direct , steric and thermal forces . as three model systems with distinctly different properties we consider the pairs barnase : barstar , cytochrome c : cytochrome c peroxidase and p53:mdm2 . in each case , proteins are modeled either as spherical particles , as dipolar spheres or as collection of several small beads with one dipole .
spherical reaction patches are placed on the model proteins according to the known experimental structures of the protein complexes . in the computer simulations ,
concentration is varied by changing box size .
encounter is defined as overlap of the reaction patches and the corresponding first passage times are recorded together with the number of unsuccessful contacts before encounter .
we find that encounter frequency scales linearly with protein concentration , thus proving that our microscopic model results in a well - defined macroscopic encounter rate . the number of unsuccessful contacts before encounter decreases with increasing encounter rate and ranges from 209000 .
for all three models , encounter rates are obtained within one order of magnitude of the experimentally measured association rates .
electrostatic steering enhances association up to 50-fold .
if diffusional encounter is dominant ( p53:mdm2 ) or similarly important as electrostatic steering ( barnase : barstar ) , then encounter rate decreases with decreasing patch radius .
more detailed modeling of protein shapes decreases encounter rates by 595 percent .
our study shows how generic principles of protein - protein association are modulated by molecular features of the systems under consideration .
moreover it allows us to assess different coarse - graining strategies for the future modelling of the dynamics of large protein complexes . |
long - range hamiltonian systems have many remarkable features @xcite , and one of them is existence of quasistationary states ( qsss ) in the way of relaxation to thermal equilibrium .
the lifetime of qsss diverges with the number of particles consisting of the system @xcite , and hence qsss are solely observable in a system with large population like self - gravitating systems @xcite .
dynamics of such a system is described by the vlasov equation , or the collisionless boltzmann equation , in the limit of large population @xcite , and the qsss , including thermal equilibrium states , are regarded as stable stationary solutions to the vlasov equation .
the system slowly goes towards thermal equilibrium with large but finite population due to finite size effects @xcite .
the qsss are observed not only in isolated systems , but also in systems under external fields .
the initial qss , which may and may not be in thermal equilibrium , is driven to another qss by the external field , and the resulting qss is not necessarily in thermal equilibrium . as a result ,
response to the external field may differ from one obtained by statistical mechanics .
indeed , in the ferromagnetic model so - called hamiltonian mean - field ( hmf ) model @xcite , the critical exponents are obtained as @xmath0 @xcite and @xmath1 @xcite with the aid of a linear @xcite and a nonlinear @xcite response theories based on the vlasov description respectively , while statistical mechanics gives @xmath2 and @xmath3 . interestingly , with another exponent @xmath4 , the non - classical exponents satisfy the classical scaling relation @xmath5 , and have universality on initial reference families of qsss in a wide class of 1d mean - field models @xcite .
the universality is derived under the assumption that the initial distribution functions depend on position and momentum only through the one - particle hamiltonian with referring to the jeans theorem @xcite .
thus , the initial states are symmetric with respect to momentum .
the symmetric initial states are also used in studies on nonequilibrium statistical mechanics @xcite , the core - halo description of qsss @xcite , nonequilibrium dynamics @xcite , and correlation and diffusion @xcite .
see also refs.@xcite .
nevertheless , asymmetric momentum distributions appear in beam - plasma systems ( see @xcite for instance ) , and are experimentally created in an ultracold plasma by optical pumping @xcite . in the hmf model ,
such distributions are stationary even asymmetric , and it is , therefore , natural to ask the response in the asymmetric case for completing the response theory . the main purpose of this article is to investigate the linear response against asymptotically constant external field around spatially homogeneous but asymmetric distributions in the hmf model .
it is worth noting that , despite its simpleness , the model shares similar dynamics with the free - electron laser @xcite and an anisotropic heisenberg model under classical spin dynamics @xcite .
the hmf model consists of plane rotators like the xy spins , and the susceptibility tensor in the hmf model is of size @xmath6 corresponding to the @xmath7- and @xmath8-directions of the rotators . for symmetric homogeneous states , the susceptibility tensor is directly diagonalized and experiences a divergence at the critical point of the second order phase transition , which is dynamically interpreted as the stability threshold of the homogeneous states @xcite .
we then ask the two questions for asymmetric momentum distributions with zero - means : is the susceptibility tensor symmetric and diagonalizable ?
does the response diverge at the stability threshold ?
we will answer these questions negatively .
the non - diagonalizable response tensor implies that the external field for @xmath7-direction induces the magnetization for @xmath8-direction , and such a response is unavoidable even changing the coordinate .
due to this non - diagonalizability , the predicted stat is not stationary , while the constant external field may drive the system to a stationary state asymptotically . in other words ,
the non - diagonalizability provides an example of discrepancy between the asymptotic states by the linear dynamics and the full vlasov dynamics .
the non - divergence of response suggests that @xmath9 and @xmath10 , and interestingly , the scaling relation @xmath11 holds , although @xmath12 might be not well defined since the spatially inhomogeneous stationary states must be symmetric by the jeans theorem @xcite .
this article is organized as follows .
the hmf model and the linear responses are reviewed in sec.[sec : model - lrt ] .
as an example of a family of asymmetric distributions , we introduce the skew - normal distributions , and investigate their stability in sec.[sec : skew - normal ] .
theoretical consequences are examined by direct numerical simulations of the vlasov equation in sec.[sec : numerics ] .
we discuss on stationarity of the predicted state in sec.[sec : stationarity - nonlinear ] .
the last section [ sec : summary ] is devoted to a summary and discussions .
the hmf model with the time - dependent external magnetic field @xmath13 is expressed by the hamiltonian @xmath14 \\ & - \sum_{j=1}^{n } [ h_{x}(t)\cos q_{j } + h_{y}(t)\sin q_{j } ] .
\end{split}\ ] ] the corresponding one - particle hamiltonian is defined on the @xmath15 space , which is @xmath16\times \mathbb{r}$ ] , as @xmath17(q , p , t ) = \dfrac{p^{2}}{2 } & - ( m_{x}+h_{x } ) \cos q - ( m_{y}+h_{y } ) \sin q \end{split}\ ] ] where the magnetization vector @xmath18 is defined by @xmath19 the one - particle distribution function @xmath20 is governed by the vlasov equation @xmath21 , f \right\ } } = 0,\ ] ] with the poisson bracket defined by @xmath22 one can straightforwardly check that any spatially homogeneous states , @xmath23 , are stationary if the external field @xmath24 is absent .
we prepare a homogeneous stable stationary state @xmath23 for @xmath25 , and add a small external field @xmath24 for @xmath26 . to avoid an artificial rotation , we require the zero - mean for @xmath23 , and consider an asymptotically constant external field accordingly . for instance , we set @xmath27 using the heaviside step function @xmath28 , and the external field drives the initial state @xmath29 to @xmath30 asymptotically .
accordingly , the one - particle hamiltonian @xmath31 $ ] changes from @xmath32 to @xmath33 , where @xmath34 and @xmath35 we introduced the averages of an observable @xmath36 with respect to @xmath29 and @xmath37 as @xmath38 it might be instructive to review the isothermal linear response to compare it with the vlasov linear response theory which will be presented in the next subsection [ sec : vlasov - linear - response ] .
the thermal equilibrium states of the hmf model are describe by the one - particle distribution functions of @xmath39 hereafter @xmath12 represents not one of the critical exponents mentioned in sec.[sec : introduction ] , but the inverse temperature .
expanding @xmath20 into the power series of @xmath40 and picking up to the linear order , we have @xmath41.\ ] ] substituting @xmath42 and @xmath43 into @xmath36 , we have the matrix formula @xmath44,\ ] ] where the correlation matrix @xmath45 is defined by @xmath46 thus , the formal solution is @xmath47^{-1 } c \begin{pmatrix } h_{x } \\ h_{y } \end{pmatrix},\ ] ] and the susceptibility tensor @xmath48 defined by @xmath49 in the limit @xmath50 is @xmath51^{-1 } c.\ ] ] divergence of @xmath52 appears at the critical point satisfying @xmath53 .
it is easy to show that the correlation matrix is now expressed by @xmath54 , where @xmath55 is the @xmath6 unit matrix . the susceptibility tensor is hence diagonalized and the diagonal elements are @xmath56 with the critical temperature @xmath57 of the second order phase transition @xcite .
the vanishing off - diagonal elements come from spatial homogeneity of @xmath23 , and symmetry of @xmath23 is not necessary .
the nonlinear response theory @xcite includes the linear response theory @xcite for symmetric @xmath23 and provides a simple expression of the linear response @xcite , but asymmetric @xmath23 is out of range .
thus , we revisit the linear response theory .
we introduce the laplace transform defined by @xmath58 the linear response theory gives the laplace transform of @xmath59 , denoted by @xmath60 , as @xmath61^{-1 } f(\omega ) \begin{pmatrix } \widehat{h}_{x}(\omega ) \\
\widehat{h}_{y}(\omega ) \\ \end{pmatrix},\ ] ] where the elements of matrix @xmath62 are @xmath63 see the appendix [ sec : appendixa ] for derivations .
the integral contour @xmath64 is the real @xmath65 axis for @xmath66 , but is continuously modified for @xmath67 to avoid the poles at @xmath68 by following the landau s procedure @xcite .
temporal evolution of @xmath69 is determined by performing the inverse laplace transform , which picks up singularities of its laplace transform .
for instance , a pole at @xmath70 gives a term having @xmath71 , which implies the landau damping for @xmath72 .
assuming that the reference @xmath23 is stable , we have no singularities on the upper half @xmath73 plane .
existence of singularities on the real axis of @xmath73 is accidental for @xmath74^{-1}f(\omega)$ ] , and we omit it .
then , the main singularity comes from the heaviside step function of the external field , whose laplace transform is @xmath75 asymptotic values of @xmath76 and @xmath77 are , therefore , obtained by picking up the pole at @xmath78 @xcite , and @xmath79 where the susceptibility tensor @xmath80 is written in a similar form with as @xmath81^{-1}f(0).\ ] ] let us rewrite the above vlasov susceptibility @xmath52 by using the dispersion function @xmath82 in the following we consider real @xmath73 which gives @xmath83 where pv represents the principal value .
the dispersion function rewrites the susceptibility as @xmath84 when @xmath23 is symmetric and hence @xmath85 , implying @xmath86 accordingly , the susceptibility tensor @xmath52 is diagonal , and the diagonal elements are @xmath87 as reported in refs.@xcite .
the susceptibility , therefore , diverges at the point @xmath88 corresponding to the stability threshold @xcite . on the other hand , when @xmath89 , the imaginary part of @xmath90 does not vanish and hence the susceptibility tensor enjoys two interesting features : ( i ) the tensor is neither symmetric nor diagonalizable by the real coordinate transformation , since the eigenvalues are not real .
( ii ) no divergence appears even at the stability threshold , since @xmath91 .
we note that , for homogeneous symmetric distributions , @xmath92 is the stability criterion and hence the divergence appears at the stability threshold .
however , @xmath92 is no more the stability criterion for the asymmetric case .
a stability criterion for the asymmetric case will be introduced in sec.[eq : nyquist - stability ] .
we introduce the skew - normal distribution for examining the linear response theory and confirming the two features mentioned in sec.[sec : vlasov - linear - response ] .
advantages of the skew - normal distribution are that it has the single peak which makes the stability criterion simpler , and that the analytically obtained mean value helps to set the total momentum zero . the density of skew - normal distribution is defined by @xmath93 where @xmath94 and @xmath95.\ ] ] the parameter @xmath96 represents the skewness , and @xmath97 results to the normal distribution .
the mean value is @xmath98 we test the homogeneous stationary states of the form @xmath99 which is normalized as @xmath100 . to set the total momentum zero , we put @xmath101 hereafter we fix the parameter @xmath102 as @xmath103
. then , the unique free parameter is the skewness @xmath96 , and the distribution is simply denoted by @xmath104 .
let @xmath105 be the unique extreme point ( the maximum point ) depending on @xmath96 .
some examples of the skew - normal distribution functions are exhibited in fig.[fig : skew - normal ] . .
@xmath106 and @xmath107 , whose maximum points are from right to left .
@xmath108 is positive ( resp .
negative ) for negative ( resp .
positive ) @xmath96.,width=302 ] for symmetric distributions @xmath23 , the formal stability criterion has been established @xcite as @xmath109 where @xmath110 is the dispersion function . to obtain the formal stability ,
@xmath23 is assumed as a function of one - particle hamiltonian , and hence we can not use this criterion for the skew - normal distributions . instead , we use the nyquist method @xcite , which was applied to asymmetric double - peak distributions in the hmf model @xcite . in our setting
, the nyquist method provides the stability criterion as @xmath111 where @xmath112 is the dispersion function and is real at @xmath113 .
see the appendix [ sec : nyquist - method ] for details .
the function @xmath114 can be rewritten as @xmath115 by performing the integration by parts and remembering @xmath116 @xcite .
the taylor expansion says that the numerator of the integrand starts from the quadratic term , @xmath117 , and hence no singularity appears in the integrand .
a rigorous treatment of the above penrose criterion is found in ref.@xcite .
the stability criterion is graphically presented in fig.[fig : nyquist ] .
the mapped real @xmath73 axis by @xmath110 intersects with the real @xmath112 axis at @xmath113 only , since @xmath118 vanishes at the unique extreme point .
consequently , we can say that the state @xmath104 is unstable iff the mapped real @xmath73 axis by @xmath110 crosses with the negative real axis on the complex @xmath112 plane . observing fig.[fig : nyquist ]
, the stability threshold of the skew - normal distributions , denoted by @xmath119 , must be in the interval @xmath120 . from symmetry with respect to @xmath96
, we have another threshold @xmath121 , and @xmath104 is stable for @xmath122 . with @xmath123
( green dotted ) , @xmath124 ( blue dashed ) and @xmath125 ( red solid ) .
each curve is the mapped real @xmath73 axis by @xmath110 , which intersects with the real @xmath112 axis at @xmath113 , the unique extreme point .
inside of the curve corresponds to the upper half @xmath73 plane.,width=302 ] the stability threshold can be estimated by precise numerical computations .
the integral in eq .
is in an infinite interval , and is impossible to perform exactly in numerics .
to estimate the infinite interval integration , we introduce the cut - off @xmath126 as @xmath127 and observe @xmath126-dependence of @xmath119 .
estimated threshold with varying @xmath126 is reported in fig.[fig : penrose ] , and is fitted by @xmath128 , where the fitting curve is obtained by the least squares method .
we hence conclude that the threshold is @xmath129 in the limit @xmath130 . with varying cut - off @xmath126 ( blue circles ) .
the black solid curve is the fitting by the least squares method in the interval @xmath131 $ ] of @xmath126 , and the red horizontal dashed line is the estimated level of @xmath132.,width=302 ]
we use the semi - lagrangian code @xcite with the time slice @xmath133 .
the @xmath15 space , the @xmath134 plane , is truncated to @xmath16\times [ -4,4]$ ] , and is divided into @xmath135 grid points .
we call @xmath136 the grid size .
the magnetization is zero for the reference homogeneous state @xmath104 , and therefore , we simply denote the response magnetization as @xmath18 instead of @xmath69
. it might be worth remarking that the truncation at @xmath137 does not conflict with the estimation of @xmath119 reported in fig.[fig : penrose ] , which requires a larger cut - off .
the reference state @xmath29 rapidly decreases as the gaussian , while the integrand in slowly decreases as @xmath138 in the large @xmath139 due to existence of the constant @xmath140 .
the obtained stability threshold is directly examined by computing temporal evolution of a perturbed state .
we prepare the perturbed initial state as @xmath141 and use @xmath142 .
temporal evolution of @xmath143 is shown in fig.[fig : stability ] , and the computed threshold @xmath119 is successfully confirmed . for the perturbed initial state @xmath144 , , with @xmath142 and @xmath145 and @xmath146 from bottom to top .
the grid size is @xmath147 .
the vertical axis is in logarithmic scale .
the stability threshold is in the interval @xmath148 , and is consistent with the estimated value @xmath129.,width=302 ] when the initial state is symmetric with respect to @xmath65 , the nonlinear response theory @xcite predicts that @xmath149 will be proportional to @xmath150 in the unstable branch .
numerical simulations captured oscillations of @xmath149 around the predicted levels and the period tends to increase as the initial state approaches to the stability threshold @xcite .
even the present asymmetric case , the scaling , oscillations and a similar tendency of periods are observed as reported in fig.[fig : responseunstable ] . as a function of @xmath151 for @xmath144 with @xmath142 .
the time window for averages is @xmath152 $ ] .
the grid size is @xmath147 .
the green straight line represents @xmath153 for guide of eyes .
the insets represent temporal evolutions of @xmath149 for the marked points .
the horizontal axis represents the scaled time @xmath154 , and the vertical axes @xmath155 and @xmath156 for the upper - left and the lower - right insets respectively.,width=302 ] we come back to the unperturbed initial distribution @xmath104 , and add the external field . from symmetry of the system we set @xmath157 without loss of generality . in order to examine the linear response theory
, we set @xmath158 to be small enough . the normalized responses @xmath159 and @xmath160 , which are susceptibilities in the limit @xmath161 ,
are reported in fig.[fig : response ] for stable states of @xmath162 and @xmath124 .
the theoretically predicted levels of responses are in good agreements with the numerical experiments in initial time intervals .
the life time of the agreements gets longer as the grid size @xmath136 increases , and is , roughly speaking , proportional to @xmath136 .
we may therefore conclude that the theoretically predicted response tensor is valid for a long time and that the non - zero off - diagonal response is observable if we use a fine grid .
( left ) and @xmath160 ( right ) with @xmath158 .
@xmath162 ( upper ) and @xmath124 ( lower ) . the grid sizes are @xmath163 ( green dotted ) , @xmath164 ( blue dashed ) and @xmath165 ( red solid ) .
the black horizontal lines are theoretical predictions : @xmath166 and @xmath167 for @xmath162 , and @xmath168 and @xmath169 for @xmath170.,width=302 ] for the whole stable region of @xmath96 , the theory is compared with numerical results in fig.[fig : responsetheory ] .
we remark that the state with @xmath97 is the thermal equilibrium state of temperature @xmath171 , and the normalized response @xmath159 coincides with the previously computed vlasov linear response @xmath172 @xcite , which is also coincides with isothermal linear response .
we stress that , as stated in the end of sec.[sec : vlasov - linear - response ] , no divergence is observed at the stability threshold , which are the left and right boundaries of the figure .
another remark is that strength of response @xmath173 for @xmath174 is greater than the symmetric case , @xmath97 .
one possible explanation for the sign of @xmath175 is as follows .
we may concentrate for @xmath176 without loss of generality . in this case
the negative part of @xmath104 is larger than the positive part around @xmath177 , and hence the small cluster being around @xmath177 induced by the external field locally has negative total momentum .
consequently , the magnetization vector turns to the negative direction of @xmath178 by the external field . .
lines are from theory .
points are from numerics with the grid size @xmath147 , and @xmath179 and @xmath180 are computed as averages over the time window @xmath181 $ ] .
diagonal element @xmath182 ( magenta solid / squares ) , and off - diagonal element @xmath175 ( black dashed / circles ) .
the region of @xmath96 is restricted in the stable interval.,width=302 ] the present non - diagonalizable susceptibility tensor comes from non - zero @xmath183 , which implies that the maximum point @xmath184 differs from the origin .
thus , we expect that asymmetric characters of the linear response tend to be hidden if the characteristic scale of @xmath65-axis , width of the separatrix , is larger than the maximum point @xmath105 , since the local total momentum in the separatrix approaches to zero . for the magnetization @xmath18 and the external field @xmath185 , the separatrix reaches to @xmath186 .
the magnetization is induced by the external field , and we have @xmath187 then , we may expect that the asymmetric characters appear for small @xmath188 satisfying @xmath189}.\ ] ] we report @xmath188 dependence of susceptibilities in fig.[fig : hdep ] for @xmath162 and @xmath124 .
the normalized responses , @xmath159 and @xmath160 , approaches to the theoretically predicted levels in @xmath190 , while the off - diagonal response , @xmath160 , goes to zero for larger @xmath188 .
dependence of susceptibilities for ( a ) @xmath162 and ( b ) @xmath170 .
open symbols are for @xmath159 , and filled symbols for @xmath160 , which are averaged over the time windows @xmath181 $ ] ( squares ) or @xmath191 $ ] ( circles ) .
the vertical black lines represent @xmath192 , and horizontal black lines the linear response levels .
the horizontal green lines are the zero level .
the grid size is @xmath147.,title="fig:",width=302 ] dependence of susceptibilities for ( a ) @xmath162 and ( b ) @xmath170 .
open symbols are for @xmath159 , and filled symbols for @xmath160 , which are averaged over the time windows @xmath181 $ ] ( squares ) or @xmath191 $ ] ( circles ) .
the vertical black lines represent @xmath192 , and horizontal black lines the linear response levels .
the horizontal green lines are the zero level .
the grid size is @xmath147.,title="fig:",width=302 ]
let us discuss a possible scenario of temporal evolution with off - diagonal response .
first of all , we show the fact that the predicted state with non - zero @xmath180 is not stationary by stating that @xmath193 and @xmath24 must be parallel in a stationary state .
jeans theorem @xcite states that an inhomogeneous distribution function is a stationary solution of the vlasov equation if and only if it depends on @xmath134 only through integrals of the one - particle hamiltonian system .
the responded state has non - zero @xmath18 and the integral is the hamiltonian @xmath194 where @xmath195 then , for a stationary state @xmath196 , we have the vanishing integral of @xmath197 since the integrand of the middle term is odd with respect to @xmath198 .
this equality and the definition of @xmath199 imply @xmath200 and we conclude @xmath193 and @xmath24 are parallel . as a result ,
the state predicted by the linear response theory is not a stationary state , and hence the system does not keep the predicted state as observed in fig.[fig : response ] .
we can point out a similarity of the present phenomenon with the nonlinear trapping @xcite . if the landau damping time scale is longer than the so - called trapping time scale , then the exponential landau damping stops and a cluster is formed by nonlinear effects @xcite . in other words ,
the state experiences the linear landau damping in an early time interval , but stops to damp by the nonlinear effects .
similarly , the state predicted by the linear response theory appears in a short time interval , and then disappears .
we conjecture that the disappearance comes from nonlinearity of the full vlasov equation .
we investigated the response tensor against an asymptotically constant external field for spatially homogeneous but asymmetric momentum distributions with zero - means by using the linear response theory .
the theory predicts two interesting characters of the susceptibility tensor : one is non - diagonalizablility , and the other is non - divergence even at the stability threshold .
the first character implies that the external field added to the @xmath7-direction induces the magnetization to the @xmath8-direction even in the simple hmf model .
the off - diagonal response is not mysterious in our setting , since anisotropy is included in asymmetry of momentum distributions . for realizing the theoretical setting
, we introduced a family of skew - normal distributions . after studying stability of the family by the nyquist method , all the theoretical consequences
are successfully confirmed by direct numerical simulations of the vlasov equation .
we stress that the crucial condition for the two characters is non - zero derivative of the reference state , @xmath201 , which never happens for symmetric @xmath23 .
one physical example of @xmath201 can be found in a beam - plasma system , whose momentum distribution consists of , for instance , a drifting maxwellian for the beam and a maxwellian for the plasma @xcite . in this example
the non - zero derivative @xmath201 is realized both with and without shifting the distribution to set the total momentum zero in general .
studying distributions having two or many peaks is a future work .
the state reached by the linear response is neither in thermal equilibrium nor in a stationary state , since the off - diagonal response is not zero , while the magnetization and the external field vectors must be parallel in a stationary state .
the life time of such a state is finite , but gets longer as the grid size becomes finer .
thus , we may expect that the off - diagonal response can be experimentally observed by using large enough number of particles .
however , non - stationarity may cause shortness of the life time comparing with the symmetric case , and revealing the time scale in which the linear response theory is valid is remained as another future work .
concerning to the above discussion , we remark on validity of the linear response theory to predict asymptotic stationary states .
we considered stable reference states , and added an external field small enough .
nevertheless , the asymptotic stationary states can not be predicted by the linear response theory for asymmetric homogeneous initial states .
analogy with the linear landau damping might be interesting , which is stopped by nonlinear effects .
recently nonlinear equations for magnetization moments has been proposed for homogeneous waterbag initial distributions in the hmf model under an external field @xcite .
an extension to non - waterbag states possibly helps to understand the nonlinear effects and to solve the puzzle on the linear response theory .
in addition to the stable initial states , perturbed unstable asymmetric initial states are also studied , and similar features are numerically observed with the symmetric case @xcite in ordering and oscillations of magnetization around the saturated states .
apart from the macroscopic variable , looking into difference in distribution functions is a remaining work .
for instance , the core - halo structure @xcite has been observed on the @xmath15 space for waterbag initial states @xcite , but it is still unclear if the present asymmetric unstable states also yield such structure in the saturated states . in this article we focused on the asymptotically constant external field corresponding to the zero total momentum , but an oscillating external field of @xmath202 is also available .
laplace transform of the external field provides poles at @xmath203 , and the denominator of susceptibility , @xmath204 , is replaced with @xmath205 as shown in , where @xmath206 is the complex conjugate of @xmath207 . as a result , setting @xmath208 where @xmath184 is the maximum point of momentum distribution ,
the susceptibility diverges at the stability threshold , which satisfies @xmath209 .
the symmetry is , therefore , not essential for the divergence of susceptibility . even in this case
, the susceptibility tensor has non - zero off - diagonal elements reflecting the asymmetry , see .
the author thanks the anonymous referees for useful comments to improve the manuscript .
he acknowledges the support of jsps kakenhi grant number 23560069 .
let @xmath210 be the hamiltonian vector field associated with the hamiltonian @xmath32 , , which is expressed as @xmath211 linearizing the vlasov equation around @xmath23 , we have the formal solution of perturbation @xmath212 as @xmath213 for the initial condition @xmath214 .
the operator @xmath215 acts on a function @xmath216 as @xmath217 where @xmath218 is the hamiltonian flow associated with @xmath32 and hence @xmath219 in our setting .
we can prove the equality @xmath220 by changing variables @xmath221 and using @xmath222 from canonical property of @xmath218 .
thus , we have @xmath223 where @xmath224 . performing the laplace transform , we obtain @xmath225 with @xmath226 \cos q \\
& - \left [ \widehat{m}_{1,y}(\omega ) + \widehat{h}_{y}(\omega ) \right ] \sin q. \end{split}\ ] ] substituting @xmath227 and @xmath228 into the linear response formula , and using the laplace transforms of @xmath229 and @xmath230 , which are respectively @xmath231 and @xmath232 we have the matrix form of @xmath233 .\ ] ] the elements of the matrix @xmath234 are exhibited in . to ensure convergence of the laplace transform
, the matrix @xmath235 is defined in the upper half @xmath73 plane .
we analytically continue the domain into the whole complex @xmath73 plane @xcite , and the resulting integral is written as @xmath236 where @xmath237 represents the principal value and is the normal integral for @xmath238 , and the second term including @xmath239 comes from the residues .
we remark that the linear response is rewritten as @xmath240 if we perform integration by parts .
the expression gives a similar form of the matrix @xmath234 with the correlation matrix @xmath241 as @xmath242 the matrix @xmath234 coincides with the correlation matrix @xmath241 as @xmath243 if @xmath23 is the maxwellian with the inverse temperature @xmath12 .
therefore , the vlasov linear response coincides with the isothermal linear response in thermal equilibrium of the homogeneous phase @xcite . in the text we concentrated on response to the external field with @xmath78 , but a general @xmath73 is also available .
the explicit form of the matrix @xmath74^{-1}f(\omega)$ ] is @xmath244^{-1}f(\omega ) = \dfrac{1}{d(\omega)\overline{d(-\overline{\omega } ) } } \begin{pmatrix } g(\omega ) & f_{xy}(\omega ) \\ - f_{xy}(\omega ) & g(\omega ) \end{pmatrix}\ ] ] where @xmath245 is the complex conjugate of @xmath73 and @xmath246f_{xx}(\omega)-[f_{xy}(\omega)]^{2}.\ ] ] in particular , the off - diagonal element is written by @xmath247 \\ & - s(\omega ) \dfrac{\pi^{2}}{2 } \left [ f'_{0}(\omega ) + f'_{0}(-\omega ) \right ] \end{split}\ ] ] and results to @xmath248 at @xmath78 as shown in the susceptibility .
if we consider the oscillating external field of @xmath202 , the susceptibility becomes @xmath249^{-1}f(\omega_{0 } ) + [ 1-f(-\omega_{0})]^{-1}f(-\omega_{0}).\ ] ] thus , for @xmath208 , where @xmath184 is the unique extreme point of @xmath23 , the diagonal elements of susceptibility diverges at the stability threshold satisfying @xmath209 . even in this case
, the oscillating external field gives the non - zero off - diagonal element as @xmath250
to review the nyquist method , we restrict ourselves in single - peak distributions including the skew - normal distributions .
let us define the set @xmath251 , where @xmath112 is the dispersion function .
if this set @xmath252 includes the origin , then there exists a root of the dispersion relation @xmath112 on the upper half @xmath73 plane , and the root corresponds to an exponential growing mode from the definition of the laplace transform .
to study the set @xmath252 , we investigate the boundary @xmath253 the boundary forms a closed curve , since @xmath254 as @xmath255 . in the limits of @xmath256 and @xmath257 , the curve approaches to @xmath258 from the positive and the negative imaginary sides respectively , since @xmath259 for @xmath260 and @xmath261 for @xmath262 , where @xmath184 is the maximum point of the single - peak distribution @xmath23 .
then , the orientation implies that the upper half @xmath73 plane is mapped onto the inside of the closed curve .
the imaginary part of @xmath112 is proportional to @xmath263 for @xmath73 real , and vanishes if and only if @xmath73 coincides with the unique extreme point @xmath184 .
thus , @xmath114 is real and @xmath264 implies that there is a root of @xmath112 on the upper half plane ( see fig.[fig : nyquist ] ) .
j. d. crawford , amplitude equations on unstable manifolds : singular behavior from neutral modes , in _ modern mathematical methods in transport theory _
( operator theory : advances and applications , vol.51 ) , w. greenberg and j. polewczak ( eds ) ( birkhuser verlag , basel , 1991 ) pp.97 - 108 . | we investigate response to an external magnetic field in the hamiltonian mean - field model , which is a paradigmatic toy model of a ferromagnetic body and consists of plane rotators like the xy spins . due to long - range interactions
, the external field drives the system to a long - lasting quasistationary state before reaching thermal equilibrium , and the susceptibility tensor obtained in the quasistationary state is predicted by a linear response theory based on the vlasov equation . for spatially homogeneous stable states ,
whose momentum distributions are asymmetric with zero - means , the theory reveals that the susceptibility tensor for an asymptotically constant external field is neither symmetric nor diagonalizable , and the predicted states are not stationary accordingly .
moreover , the tensor has no divergence even at the stability threshold .
these theoretical findings are confirmed by direct numerical simulations of the vlasov equation for the skew - normal distribution functions . |
supported graphene samples have been prepared by mechanical exfoliation of a hopg crystal ( momentive performance ) onto a standard 90 nm sio@xmath40si ( graphene supermarket ) wafer under ambient conditions .
graphene was grown on 25 @xmath1 m thin copper foils ( 99,8% , alfa aesar gmbh & co kg , karlsruhe ) using low - pressure chemical vapour deposition @xcite . before growth
the copper foils were rinsed with acetone then placed in acetic acid ( p.a . ) for 10 min and finally cleaned with water , acetone and isopropanol .
then the foils were dried under a nitrogen stream , put into a quartz tube and introduced into a furnace ( gero f40 - 200 , neuhausen ) .
pumping the tube to @xmath41 mbar was followed by setting a h@xmath28-flow of 50 sccm and subsequent heating to 1288 k in two steps ( 500k / h up to 1223 k , 100k / h up to 1288 k ) . after 3 h the h@xmath28-flow was reduced to 10 sccm and a methane flow of 70 sccm was introduced for 15 min .
then the samples were cooled down to room temperature under the ar / h@xmath28 atmosphere .
the grown graphene sheets were transferred onto tem grids . to this end , poly(methyl methacrylate ) was spun in two steps onto the graphene to stabilize it during the transfer process .
first , a layer of low molecular weight pmma ( 50k ) and then a second layer of high molecular weight pmma ( 950k ) was spin - cast for 30 s at 4000 rpm onto the graphene .
then the back site of the copper foil was cleaned using o@xmath28 plasma etching for 60 s , followed by etching away the copper using an ammonium persulfate solution ( 6.5 g ammonium persulfate/100 ml water ) over night . after placing the graphene onto the quantifoil tem grid ,
the pmma was dissolved in a critical point dryer . to additionally clean graphene the samples were introduced into ultra - high vacuum ( @xmath42 mbar ) and annealed for 2 h at 350 @xmath31c .
samples haven been transferred to the different accelerators for irradiation . in the high energy regime ( @xmath43 mev / u )
, ion irradiation leads to ionization and excitation of target atoms .
the most important parameter to achieve a given modification is thus the electronic stopping power @xmath6 .
this deposited energy per track length depends on the projectiles mass , its kinetic energy , and the target properties ( mass , stoichiometry , density ) .
thus , with a given sample material , the desired stopping power can be achieved by simply tuning the kinetic energy of the projectile .
this again requires accelerators which operate with many different ions at the various beam times . in the case of graphene
, a given modification will be achieved as long as a certain threshold is overcome ( see main text ) independent on the type of ion that was used for the respective irradiation . for simplicity and easier comparison
we have therefore calculated the stopping power in graphite with the standard software code srim @xcite and used these numbers throughout the manuscript instead of naming the specific ion type and its energy .
samples have been irradiated with various ions ( see table below ) at the irrsud beamline of the ganil ( caen , france ) and at the tandem van de graaff accelerator at the rbi ( zagreb , croatia ) which are both equipped with a positioning system for glancing incidence irradiation .
irradiation took place under vacuum conditions with @xmath44 mbar . + + [ cols="^,^,^,^",options="header " , ] + + after irradiation samples have been analyzed by afm , raman and ac - hrtem .
the afm measurements were performed using a veeco dimension 3100 afm in tapping mode .
the cantilevers used for these measurements are nanosensors nchr with a typical resonant frequency of 300 khz . as absolute height measurements of 2d materials in tapping mode
are difficult to obtain , a typical height scale for our samples is shown only in fig . [ 1 ] and in fig . [ 5 ] , where large variations due to the substrate modifications occur . for the rest of the afm images the height scale
is given in the figure captions instead .
raman spectra were taken with a renishaw invia raman microscope .
the laser excitation wave length was 532 nm and the laser intensity was always kept below 0.4 mw with a spot size below 1 @xmath1 m to avoid heating effects .
+ + the high - resolution transmission electron microscopy was conducted using an fei titan 80 - 300 microscope equipped with an image side hexapole spherical aberration corrector and a field emission gun .
the extraction voltage of the gun was lowered to 2 kv to reduce the energy spread of the beam .
the spherical aberration was tuned to @xmath45 m and the imaging was conducted at scherzer focus , resulting in dark atom contrast .
financial support from the dfg within the spp 1459 `` graphene '' is gratefully acknowledged .
mk and mj acknowledge the croatian science foundation ( pr.no .
8127 ) which supported this investigation .
support by the croatian centre of excellence for advanced materials and sensing devices is also acknowledged .
we thank v. freie soler for support with sample preparation .
10 q. wu , y. wu , y. hao , j. geng , m. charltion , s. chen , y. ren , h. ju , h. li , d. w. boukhvalov , r. d. piner , c. w. bielawski , and r. s. ruoff .
selective surface functionalization at regions of high local curvature in graphene . , 49:677 , 2013 .
t. hallam , a. shakouri , e. poliani , a. p. rooney , i. ivanov , a. potie , h. k. taylor , m. bonn , d. turchinovich , s. j. haigh , j. maultzsch , and g. s. duesberg .
controlled folding of graphene : grafold printing .
, 15:857 , 2014 .
papalo , r. thomaz , l.i .
gutierres , v.m .
de menezes , d. severin , c. trautmann , d. tramontina , e.m .
bringa , and p.l .
confinement effects of ion tracks in ultrathin polymer films .
, 114:118302 , 2015 .
r. ritter , r. a. wilhelm , m. stoeger - pollach , r. heller , a. muecklich , u. werner , h. vieker , a. beyer , s. facsko , a. goelzhaeuser , and f. aumayr .
fabrication of nano pores in 1 nm thick carbon nanomembranes with slow highly charged ions .
, 102:063112 , 2013 .
x. li , w. cai , j. an , s. kim , j. nah , d. yang , r. piner , a. velamakanni , i. jung , e. tutuc , s.k .
banerjee , l. colombo , and r.s .
large - area synthesis of high - quality and uniform graphene films on copper foils .
, 324:1312 , 2009 .
m. woszczyna , a. winter , m. grothe , a. willunat , s. wundrack , r.stosch , t. weimann , f. ahlers , and a. turchanin .
all - carbon vertical van der waals heterostructures : non - destructive functionalization of graphene for electronic applications .
, 26:4831 , 2014 . | the ability to manufacture tailored graphene nanostructures is a key factor to fully exploit its enormous technological potential .
we have investigated nanostructures created in graphene by swift heavy ion induced folding . for our experiments , single layers of graphene exfoliated on various substrates and freestanding graphene have been irradiated and analyzed by atomic force and high resolution transmission electron microscopy as well as raman spectroscopy . we show that the dense electronic excitation in the wake of the traversing ion yields characteristic nanostructures each of which may be fabricated by choosing the proper irradiation conditions .
these nanostructures include unique morphologies such as closed bilayer edges with a given chirality or nanopores within supported as well as freestanding graphene .
the length and orientation of the nanopore , and thus of the associated closed bilayer edge , may be simply controlled by the direction of the incoming ion beam . in freestanding graphene
, swift heavy ion irradiation induces extremely small openings , offering the possibility to perforate graphene membranes in a controlled way . *
introduction * + folding is an old an every - day technique which can be used to shape a material but also to modify material properties such as mechanical strength . while folding is usually applied in the macroscopic world , recent work demonstrated that it might be useful also for nanoscaled materials such as graphene @xcite , the most prominent of the two - dimensional materials . in the case of graphene
, folding has attracted quite a lot of attention in particular , as it offers the chance to tune its mechanical , chemical , and electronic properties @xcite .
for example , graphene folds locally strengthen the material @xcite , they may be used as transport channels , and graphene origami boxes could serve as containers for hydrogen storage @xcite for which then an electric field can be used to unfold graphene and release the stored gas .
at the point where folded graphene is bent , a closed bilayer edge is formed . in these closed bilayer edges
the covalent bonds are bent , which effectively changes the local chemical reactivity and transport properties @xcite . as a consequence a strong magneto - photoelectric effect @xcite , bandgap openings @xcite , and enhanced spin - orbit interaction @xcite are expected .
furthermore , folding of graphene is accompanied by the formation of bilayer graphene , either in the regular stacking or in its twisted form .
the latter shows rotational angle dependent optical properties @xcite , flat bands close to the fermi energy @xcite , chiral tunneling @xcite , and large interlayer resistivity , strongly dependent on temperature @xcite .
this huge scientific and technological potential of folded graphene may however only be exploited if reliable methods for controlled folding are at hand .
folding of graphene has so far been achieved by ultrasound sonication @xcite , the use of chemicals @xcite , by using the tip of an atomic force microscope ( afm ) @xcite , femto - second laser ablation @xcite , and swift heavy ion irradiation ( shi ) @xcite .
most the aforementioned methods create foldings in a more or less random way with respect to size and orientation , as well as the number of foldings .
in addition , they require edges , i.e. folding of graphene without a pre - existing defect can not be achieved .
the irradiation with swift heavy ions is a notable exception and may in principle overcome all these disadvantages , because type of ion , energy , angle of incidence and fluence are easy to control parameters , and in addition , shi may induce foldings anywhere in a graphene flake .
in general , this type of ion beams represent a powerful processing tool to produce nanoscaled material modifications which are unachievable by other methods @xcite . due to their high kinetic energy , typically several hundred kev per nucleon and above ,
the main interaction with matter is not via nuclear collisions , but via dense electronic excitation .
thereby extreme energy densities are achieved and the affected area is centred around the trajectory of each ion . in a grazing incidence geometry
the resulting material changes at the surface may extend up to several microns .
thus , nanostructuring of graphene by shi seems perfectly feasible .
however , there exists a profound lack of knowledge as so far neither the mechanism nor the potential of this approach have been investigated yet . to assess the potential and to determine the conditions for nanostructuring graphene , we have prepared and irradiated supported and freestanding graphene samples with swift heavy ions .
we have analysed the resulting morphological changes with optical spectroscopy and various microscopy methods down to the atomic scale and determined the influence of the ion energy .
finally , we varied the experimental conditions to investigate the influence of the substrate and graphene itself on the irradiation - induced graphene nanostructures . this paper offers evidence that shi irradiation can be used to perforate supported and even freestanding graphene .
it provides the necessary threshold energy per track length for nanostructuring and the details on how to obtain certain types of graphene nanostructures .
thus , the full potential of shi irradiation in conjunction with the novel material class of two - dimensional solids is revealed . + * results and discussion * + to investigate nanostructuring of graphene by shi induced folding , the ion beam is directed towards the sample as schematically depicted in fig .
[ 1](a ) . for a first experiment
, graphene has been exfoliated on a pre - patterned sio@xmath0 substrate ( 3 @xmath1 m diameter holes , 8 @xmath1 m in depth written by photolithography and consecutive dry etching ) . a typical , high quality single layer ( see raman spectrum in fig .
[ 1](c ) ) graphene flake on such a patterned substrate is shown in the optical image in fig . [ 1](b ) .
the red box in fig . [ 1](b ) marks the area where the afm image has been taken which shows suspended graphene in the upper left corner and sio@xmath0 supported graphene elsewhere , see fig . [ 1](d ) .
substrate .
( c ) raman spectra taken on the pristine ( orange ) and on the irradiated graphene sheet ( blue ) , here with 91 mev xe , @xmath2 .
a distinct @xmath3-band at 1350 @xmath4 is observed .
( d ) afm topography of the irradiated graphene flake within the red box in the optical image ; ion beam direction in ( d ) marked by the white arrow .
note , that the edge of the few layer graphene ( flg ) in the upper right corner runs parallel to a low - indexed direction of the single layer underneath , which was used to determine @xmath5 for this sample .
( e ) nanoscaled slit - pore in the suspended graphene sheet .
their length is about one tenth of the length of ( f ) the opening found in the substrate supported graphene . here
, the folded bilayer is clearly seen.,width=529 ] this sample has been irradiated with 91 mev xe ions at an angle of incidence @xmath2 with respect to the surface .
the irradiation induced defects show up as a notable @xmath3 peak in the raman spectrum .
atomic force microscopy images taken from graphene supported by the sio@xmath0 substrate reveal that these defects are extended foldings which are in average one micron long ( fig .
[ 1](d , f ) ) and are created with an efficiency , i.e. the number of foldings per ion , of 1 . at first glance ,
no foldings or modifications at all can be observed in the suspended graphene sheet in fig . [ 1](d ) .
however , the zoom - in in fig .
[ 1](e ) shows that nanoscaled slit pores appear in suspended graphene .
the length of these slit pores is only about one tenth of the length of the foldings on the sio@xmath0 substrate .
fig .
[ 1 ] presents the three characteristic irradiation induced graphene nanostructures , ( twisted ) bilayers , closed bilayer edges , and pores in supported graphene as well as nanopores in freestanding graphene .
note , that foldings and pores can be created directly within the flake and that no open edges are required as with other methods .
we will proceed by discussing the conditions under which each of these structures might be fabricated and how these can be met .
an important issue is certainly the minimum deposited energy per track length that is required to obtain a given modification . in the energy regime of swift heavy ions ,
typically projectiles with 0.5 to a few mev per atomic mass unit , the deposited energy will depend on the kinetic energy of the ion and its mass .
the deposited energy per path length @xmath6 is usually given in kev / nm and is called electronic stopping power @xmath7 , because the primary energy transfer channel involves electronic excitations and ionization of the target material .
the subsequent processes may involve non - thermal and thermal transitions such as ultrafast melting and coulomb explosion on fs time scales @xcite , and melting on the ps time scale due to electron phonon coupling ( thermal spike ) @xcite . in order to determine the threshold for material modification to occur in terms of stopping power , graphene on sio@xmath0 samples
have been irradiated with shi of different stopping powers in the range from 2 18 kev / nm ( by choosing different ions at different kinetic energies , see methods ) .
as an example we show in fig .
[ 2 ] graphene samples irradiated with 14 kev / nm , 6 kev / nm , and 4.3 kev / nm , all with the same nominal incidence angle of @xmath8 .
note , that we took great care to determine the accurate incidence angle in this experiment ( see @xcite for details ) so that any deviation between these samples due to a difference in the incidence angle @xmath9 can be excluded . comparing the afm measurements in fig .
[ 2](a)-(c ) it can be clearly seen , that the size of foldings ( and the corresponding pores , respectively ) in graphene depends on the stopping power : while the foldings with 6 kev / nm are typically 60 @xmath10 20 nm in length , foldings created with 14 kev / nm are twice as large ( 130 @xmath10 40 nm ) .
note that the scatter in length maybe due to the statistical nature of the electronic stopping and is expected to be even larger in the vicinity of a threshold .
furthermore , fig .
[ 2](c ) shows that for lower stopping powers not only the size of the foldings is affected , but also the folding efficiency approaches zero .
we found that the minimum stopping power down to which foldings can be observed is @xmath11 kev / nm , marked by the red solid line in fig . [ 2](d ) .
this threshold is somewhat higher than the one recently reported for perpendicular irradiation @xcite and lower than the threshold for defect creation in graphite @xcite where foldings have never been observed .
the surprisingly low value offers the technical advantage that also smaller accelerators can be used for folding of graphene . .
( a ) afm topography of slg on sio@xmath0 irradiated with 84 mev ta ions yielding @xmath12 kev / nm , ( b ) irradiated with 23 mev i yielding @xmath13 kev / nm , and ( c ) irradiated with 15 mev si yielding @xmath14 kev / nm .
height scales are 0 to 5 nm .
( d ) analysis of @xmath15 ( black dots ) and @xmath16 ( blue squares ) intensity ratios from raman spectra .
black line to guide the eye .
the red line represents the minimum stopping power that induces foldings detectable by afm.,width=566 ] the threshold for folding to occur can be compared to the threshold for defect creation as determined from raman spectra taken at an excitation wave length of 532 nm from samples irradiated with ions of varying stopping powers but at a constant fluence , here 5@xmath1710@xmath18 ion/@xmath1m@xmath19 under perpendicular incidence . the plot in fig . [ 2](d ) shows a steady increase of the i@xmath20/i@xmath21 ratio ( left scale , black dots ) with increasing ion stopping power .
apart from the i@xmath20/i@xmath21 ratio which quantitatively describes the size of the defects , the i@xmath20/i@xmath22 ratio can be used to investigate the nature of defects @xcite .
eckmann et al .
found a maximum for the i@xmath20/i@xmath22 ratio ( @xmath23 13 ) for defects due to @xmath24 hybridization of carbon with other atoms .
lower i@xmath20/i@xmath22 ratios are indicative of vacancy like ( @xmath23 7 ) and boundary like defects ( @xmath23 3.5 ) .
note , that in fig .
[ 2](d ) above the threshold stopping power for foldings ( red line ) the i@xmath20/i@xmath22 ratio ( right scale , blue squares ) is between 7 and 9 which corresponds to vacancy like defects . below the threshold stopping power for foldings ( as determined by afm )
an increase of the i@xmath20/i@xmath22 ratio to 12 14 can be observed which corresponds to @xmath24 hybridization .
this indicates that not only the size of the structurally - disordered area but also the nature of defects can be controlled by the shi stopping power .
furthermore it points towards a number of aligned vacancy defects acting as the artificial point of failure necessary for folding , while @xmath24 defects are not as effective .
this is in full agreement with the reported mechanical properties of defective graphene . in terms of stiffness and intrinsic strength
, @xmath24 defective graphene shows only a slightly reduced breaking strength while a significant degradation of mechanical properties was observed in the vacancy regime @xcite .
+ in the following we will discuss in more detail the nanostructures presented in fig . [ 1 ] , and how they may be fabricated in a controlled manner .
we will begin with the nanoscaled slit pores in freestanding graphene .
it is quite difficult to image suspended graphene with afm in tapping mode as the oscillating tip in this operation mode comes into contact with the membrane which easily bends due to the applied force .
this results in a blurred topography image offering only little details , see fig .
[ 1](e ) . to circumvent this problem and study the ion induced modification with the highest possible resolution , we investigated the samples by aberration - corrected high - resolution transmission electron microscopy ( ac - hrtem ) . for the ac - hrtem measurements ,
graphene grown on a copper foil was transferred using polymethylmethacrylate ( pmma ) onto a quantifoil - tem grid which is shown in fig .
[ 3](a ) .
the suspended graphene sheets are located on top of the holes with a diameter of 2 3 @xmath1 m which is sufficiently large for the ac - hrtem measurement .
kev / nm ) .
( a ) optical image of a tem grid onto which cvd graphene has been transferred .
( b ) ac - hrtem measurement of the tem - grid suspended graphene after irradiation with shi . here
, contaminants can be seen as dark areas on the graphene flake which are due to the transfer process and adsorbates / hydrocarbons from the air @xcite . for comparison
we show again and in the same scale the afm image from fig.1 , where the typical morphology after grazing incidence irradiation of freestanding graphene is a slit pore .
( d ) zoom - in of the ac - hrtem measurement reveals that nanoscaled rifts from the afm image are in fact foldings .
( e ) atomically resolved image shows twisted bilayer ( orange and gray ) and closed bilayer egde ( red).,width=566 ] to compare the results from afm ( fig . [ 1](c ) ) with the ac - hrtem measurements , both images are shown next to each other and using the same scale in fig .
[ 3](b , c ) .
most notably , the slit pores seen in the afm image are revealed by the ac - hrtem measurement to be miniature foldings ( fig . [ 3](d ) ) , with an aspect ratio ( length to width ) of approximately 2 to 1 .
this zoom - in also reveals that surprisingly , not too many carbon atoms are missing as folding back the orange areas ( where the bilayer could be roughly identified ) would again close the nanopore .
an atomically resolved image of such a folding clearly reveals the unique nature of the nanostructure created by the ion impact , see fig . [
3](e ) . here , the unaffected single - layer graphene area is colored in violet , while the border of the folded part is marked orange to outline the extension of the bilayer .
the line along which graphene has been flipped over and the closed bilayer edge has formed , is highlighted in red .
the high - resolution ac - hrtem image reveals a straight edge at the semicircular bend with no missing carbon atoms , corroborating the hypothesis that the zipped - up graphene layer flips over and attaches itself to the unaffected material underneath ( or above ) , eventually trapping residual atoms and contaminants in the process . despite the disruptive nature of the process ,
the folded areas maintain to a large extent a well - ordered graphene structure , resulting in twisted bilayer sheets .
this becomes evident from the distinct moir pattern ( areas coloured in orange ) which is due to an orientational lattice mismatch between the first and second layer graphene of @xmath25 in this case .
areas on the bilayer colored in gray are contaminated which might play a role for the unusual and differing heights typically found in afm images .
finally , we wish to point out , that typically the major part of the boundary of a nanopore consists of an intact closed bilayer edge .
also , it is safe to assume that the aspect ratio of the nanopores can be controlled by the incidence angle as is the case for supported graphene .
this fact , together with the unique and well - defined edge configuration , might prove advantageous for the fabrication of porous graphene membranes for filtering applications where pore size and the termination of the pore edge play an important role for filter performance @xcite .
+ if non - twisted , twisted bilayers , or closed bilayer edges are to be fabricated in supported graphene samples , the most important parameters will be the two angles introduced in fig .
[ 1](a ) : the polar angle @xmath9 which denotes the angle of incidence of the ion with respect to the sample surface and the azimuthal angle @xmath26 , the angle between the incident ion and a given crystallographic edge of the 2d - crystal ( which may be either armchair or zigzag ) .
although irradiation under perpendicular incidence can be used to induce defects in graphene and graphene related materials , as has been shown for singly charged ions @xcite , highly charged ions @xcite and swift heavy ions @xcite , no foldings are introduced under these conditions . however , under glancing incidence angles individual shi do cause foldings of different length and shape , for which a two - step model has been proposed @xcite . in the first step the shi irradiation causes via electronic excitations an extended line of defects ,
e.g. in form of a transition from hexagonal carbon rings to pentagonal and heptagonal ones .
subsequent to the creation of these primary defects , the hillocks which are formed by the shi interaction with the substrate surface , are pushing through the graphene layer which is then unzipped along the ion trajectory .
this hypothesis implies that the glancing incidence condition is a prerequisite for the folding as it yields a significant number of defects all aligned in a row to sufficiently weaken the graphene layer .
therefore , we have determined the minimum polar angle which still allows for a successful folding process by irradiating a set of sio@xmath0 supported graphene samples at a constant fluence of 15 ions/@xmath1m@xmath19 and varying incidence angles @xmath9 .
exemplary results of this irradiation series shown in fig . [ 4](a)-(c ) , make it evident that extremely long ( in the range of microns ) closed bilayer edges can be fabricated , while the width of the associated bilayer is on the order of a few 10 nanometres and thus close to what would typically be achieved with lithographic techniques @xcite . at large angles of incidence , e.g. @xmath27 as shown in fig .
[ 4](c ) , the irradiation does not yield detectable foldings anymore , instead sub-10 nm pores are observed .
detectable folding by means of ambient afm occurs in graphene on sio@xmath28 up to incidence angles of @xmath29 . for comparison we like to point out that irradiation of graphene on ir(111 ) with kev ions at comparable angles of incidence yields chains of vacancy clusters @xcite . in that energy regime ,
the primary energy dissipation takes place via atomic collisions and thus , the occurrence of nanopores framed by foldings is a unique consequence of the dense electronic excitations triggered by the shi . .
( a ) afm image of foldings after irradiation with @xmath30 kev / nm at @xmath2 .
height scales are 0 to 5 nm . for this ultra grazing incidence angle , foldings are mainly oriented along the shi trajectory ( dashed orange line ) as the corresponding polar diagram shows .
the error of @xmath26 is in the range of @xmath10 3@xmath31 mainly because of piezo drift during afm measurement .
( b ) for less grazing incidence angle ( here @xmath32 ) foldings are shorter and are mainly oriented close to low - indexed directions of the graphene lattice .
( c ) at large incidence angles such as @xmath27 , nanopores with a diameter @xmath33 nm are observed by afm.,width=585 ] controlling the length of graphene nanostructures with closed bilayer edges is only one factor . to actually make use of twisted bilayer and closed bilayer edges , with their specific electronic and chemical properties , it is absolutely necessary to control the lattice mismatch or the orientation of the closed bilayer edge , respectively .
it is well known that in mechanically exfoliated graphene samples the angles between edges are typically given in multiples of 30@xmath34 .
this indicates that graphene is cleaved preferably along its low - indexed crystalline directions and either zigzag or armchair edges are generated @xcite .
for the same reason , artificially induced foldings of graphene tend to exhibit folding angles close to 30@xmath34 as well @xcite . with shi irradiation
this limitation can be overcome . to demonstrate this
, we have analyzed afm images such as figs .
[ 4](a),(b ) by measuring the azimuthal angle @xmath26 of the closed bilayer edge with respect to a given single layer graphene edge of the same flake which may be either armchair or zigzag . in fig .
[ 4](a),(b ) we show the corresponding angle distributions determined for two different incidence angles @xmath9 . for an ultra glancing incidence angle like @xmath9=0.1@xmath34 , the orientation of the closed bilayer edges does not show the characteristic multiples of 30@xmath34 ,
instead @xmath26 is almost exactly oriented along the direction of the impinging shi as marked by the dashed orange line in the polar diagram of fig .
[ 4](a ) . in fig .
[ 4](b ) the incidence angle was increased to @xmath9=0.8@xmath34 and the resulting closed bilayer edges show a @xmath26-distribution which is no longer aligned with the direction of the impinging shi , but is centered around the 60@xmath34 and 90@xmath34 directions , corresponding to the graphene zigzag and armchair orientations ( or vice versa ) . going to even higher incidence angles results in a widespread distribution of @xmath26 with no obvious correlation to the ion beam direction ( data not shown ) . in conclusion
, it can be said that we are able to obtain graphene nanostructures with closed bilayer edges oriented exactly along the shi trajectory with an efficiency of more than 90% at low incidence angles like @xmath2 . only for higher incidence angles
, the orientation of the closed bilayer edge typically exhibits multiples of 30@xmath34 , i.e. they tend to be oriented along the crystallographic lattice of graphene .
+ finally , we wish to extend the discussion beyond sio@xmath28 as a substrate because folding by shi can be achieved in graphene supported on virtually any substrate . from the huge difference in size and length for supported and suspended
graphene ( see fig . [ 1 ] ) one can already conclude that the substrate does play an important role for the folding .
furthermore , whenever graphene is exfoliated or transferred onto its supporting substrate material , an interfacial layer will be present after preparation .
this layer will consist mostly of water and will significantly influence the physical properties of graphene .
for example , it can lead to blocking of charge transfer from the substrate to graphene @xcite or reduce the adhesion forces between graphene and the substrate @xcite . upon irradiation
the shi will interact with the substrate and the interfacial layer along its trajectory and the intensive and abrupt heating of the substrate material as well as the trapped water layer can lead to a significant ejection of particles which will certainly affect the folding process .
kev / nm , @xmath9=1.5@xmath35 ) on different substrates . for better comparison
all images are shown in the same size but with different height scales shown to the left of each image .
( a ) suspended graphene , ( b ) graphene on sio@xmath28 , ( c ) graphene on sic , and ( d ) graphene on pmma.,width=566 ] to demonstrate the role of the substrate more clearly , we have compiled data obtained from various substrate supported graphene samples as exemplary shown in fig .
[ 5 ] . here , the incidence angle has been kept constant at @xmath36 . in suspended graphene ( fig . [ 5](a ) ) , the irradiation does create defects which appear as holes rather than foldings in these images . however , as we have just shown , they are indeed foldings and for the sake of clarity we will stick to the term . with a typical length of 20 30 nm ( averaging over 10 foldings )
the foldings in suspended graphene are much shorter than on any of the substrates .
graphene on sio@xmath0 ( fig .
[ 5](b ) ) shows foldings with a length of 150 nm on average which is about five times the value of suspended graphene .
note , that stopping power and angle of incidence were kept constant here , but that the different substrates react quite differently to the irradiation .
the sio@xmath0 substrate shows faint surface tracks after shi irradiation , while sic just shows very shallow depressions because only the topmost si atoms are sublimating upon irradiation @xcite . nevertheless , the foldings in graphene on sic ( fig . [ 5](c ) ) are almost of the same length as in graphene on sio@xmath0 . with the radiation sensitive polymer pmma as substrate ( fig . [ 5](d ) ) , the foldings are more than twice as long ( around 370 nm ) as the foldings on sio@xmath0 and their length exceeds the length of the foldings in freestanding graphene by as much as a factor of 15 .
in addition , the foldings are also broader .
this nicely demonstrates that apart from the interfacial water layer , the amount of material ejected from the substrate due to the interaction with the ion influences the size of the folding to a large extent . for the most common conditions used in current basic research , i.e. a sio@xmath28 substrate with a water layer present , very good results with respect to folding size and efficiency are easily obtained .
in freestanding graphene , where both substrate and interfacial layer are absent , nanopores with a typical width of 10 20 nm can be fabricated .
this , together with the fact that the shape of the nanopore can be tailored by various parameters proves that shi are indeed a powerful tool to perforate supported and free - standing graphene .
nanoporous graphene has been suggested as a key factor in many applications .
for example , graphene nanomeshes supported on sio@xmath28 have recently been shown to be semiconducting thus offering new possibilities for bio - sensors and spintronic devices @xcite .
graphene perforated by shi may not yield the inherent periodicity of nanomeshes but as an advantage the pore density is much easier to control .
swift heavy ions represent the standard technique to fabricate track etched membranes @xcite and based on our results we propose that the concept can be extended to graphene - based membranes which promise an unprecedented level of transport rates ( permeability ) in filtering applications as the quasi two - dimensional selective membrane would exhibit negligible wall interactions @xcite .
+ + * summary * + in summary , we could show that graphene is very sensitive to ionizing particle radiation . as a consequence
, swift heavy ion beams can be used as a versatile tool to intentionally fold graphene , yielding nanopores and graphene nanostructures with specific characteristics such as size , orientation or shape . in general
, the technique offers a high level of control as ion beam parameters are easily changed , controlled and maintained . to achieve perforation and folding of graphene , there are two conditions to be met .
first , the ion has to deposit sufficient energy per track length to induce vacancy like defects in the graphene sheet ( @xmath37 5 kev / nm ) .
this can be readily achieved with small scale accelerators as they are commonly used in standard ion beam facilities .
second , graphene has to be irradiated under non - perpendicular angles of incidence .
this poses no severe limitation as in supported graphene on sio@xmath28 e.g. , foldings occur already at angles as large as @xmath38 45@xmath34 .
however , for tailoring specific nanostructures , additional criteria have to be fulfilled . while the length of the folding are mainly determined by the incidence angle @xmath9 and the energy loss s@xmath39 , our comprehensive experimental studies show that width and shape of the folding can be tuned in a wide regime by the appropriate choice of the substrate material .
of all two - dimensional materials graphene will probably be the one which is the least susceptible to effects of ionizing irradiation .
therefore , we are confident that swift heavy ions are a useful tool to fabricate similar nanostructures in other 2d materials such as mos@xmath28 and hexagonal bn , or even heterostructures thereof . |
rogue waves ( rws ) appear in variety of physical systems , such as in oceans @xcite , in atmosphere @xcite , in optics @xcite , in plasma @xcite .
a fundamental property of such extreme waves is that they appear from nowhere , reach heights up to twice the height of surrounding waves , and disappear without a trace @xcite .
significant theoretical progress has been achieved in recent years in order to describe the formation of rws within the framework in weakly nonlinear evolution equations @xcite .
in fact , rws may be modeled in a general context by exact breather solutions of the nonlinear schrdinger equation ( nls ) @xcite , which describe the propagation of stationary and pulsating wave packet dynamics in nonlinear dispersive media .
an appropriate model to describe strong localizations in the medium of interest is the family of doubly - localized breather solutions , also referred to as akhmediev - peregrine ( ap ) solutions .
the first - order solution is known as the peregrine breather ( pb ) solution @xcite ; it is localized in space and in time , since it is the limiting case of time - periodic kuznetsov - ma solitons @xcite as well as space - periodic akhmediev breathers @xcite , when the period of these periodic solutions tends to infinity .
furthermore , the ap amplifies the amplitudes of the carrier by a factor of three and higher .
pb dynamics has been validated experimentally in optics @xcite , in water waves @xcite and in plasma @xcite .
more recently higher - order breathers have been observed as well @xcite .
there are indeed several interesting rw pattern solutions which the nls may admit .
fundamental patterns may consist of a simple central high peak , surrounded by several gradually decreasing solitonic peaks , an equal height triangular pattern , or a circular pattern @xcite .
doubly - localized ap solutions are symmetric with respect to both the spatial and the temporal co - ordinate in theory
. however , several non - symmetrical rogue waves under fundamental pattern have been observed in water tank , see for instance figs .
4 - 6 in @xcite , figs . 3 , 4 and 6 in @xcite ,
figs . 3 - 9 in @xcite .
the observed non - symmetry of wave profiles in the wave flume is due to higher - dispersive and to the effect of the mean flow , not captured in the nls , which can be explained by more accurate evolution equation , such as the generalized nls in optics and the modified nls ( mnls ) in water waves @xcite . independently from these higher - order effects , it is an interesting task to derive doubly - localized and non - symmetric solutions of the nls .
physically , non - symmetric structures have different and fundamental distribution of energy and would be of interest for several physical applications . in this work ,
we derive exact solutions of the manakov system up to third - order and as a special class corresponding family of non - symmetric doubly localized solutions of the nls .
in addition , we demonstrate the physical accuracy of the derived solutions to describe strong localizations in water waves by generating experimentally the derived second - order solution in a water wave tank .
it is shown that the expected amplitude amplification of the background is reached .
moreover , the experimental results are in very good agreement with numerical mnls simulations . this paper is organized as follows . in section 2
, we introduce the lax pair of the manakov system and solve the lax pair equations , with spectral parameter @xmath0 , from a periodic _ seed _ solution .
in other words , we derive eigenfunctions of the manakov system , associated with the spectral parameter @xmath0 . in section 3
, we give a determinant representation of the @xmath1th - order vector rws of the manakov system . in particular , exact expressions for the first- , second- and third - order rw solutions of the nls equation , localized in both time and space are reported .
the latter solutions have the property to be non - symmetric with respect to the spatial co - ordinate .
furthermore , the differences between the non - symmetric and corresponding symmetric doubly - localized rws of the nls equation are discussed in detail . in section 4 , we shall compare experimental data , related to the evolution of the non - symmetric solution , derived in this work , in a water wave tank , with nls maximal wave profile predictions as well as with numerical mnls simulations .
finally , we summarize the main results in section 5 .
recently , rw solutions of the multi - component soliton equations have been derived .
these solutions attracted the scientific interest @xcite .
the manakov system @xcite is an important integrable two - component nls system and can be applied to optical as well as hydrodynamic systems @xcite . in particular ,
if @xmath2 is a solution of the manakov system @xmath3 then @xmath4 is a solution of the nls equation @xmath5 in the present paper , we shall construct the solutions of the manakov system under the condition @xmath6 by the darboux transformation ( dt ) from a periodic _ seed _ solution , which imply several non - symmetrical extreme localizations of the nls equation .
the dt is a powerful method to generate soliton @xcite , breather @xcite , rw @xcite solutions of the nls equation .
the breather solution of the manakov system @xcite has been constructed by the dt from a periodic _ seed _ solution .
a crucial step of the dt method is to find the solutions , i.e. eigenfunctions of the linear partial differential equations , defined by the lax pair of the soliton equations .
therefore , we shall first solve the eigenfunctions of the manakov system in this section in order to provide necessary preparatory work for the dt in next section .
the manakov system is produced by the compatibility of the associated lax pair equations : @xmath7 where @xmath8 here , @xmath9 is a vector eigenfunction associated with the spectral parameter @xmath0 .
+ to obtain rw solutions of the manakov system , we start with periodic _ seed _ solutions , @xmath10 where @xmath11 .
then , the dispersion relations can be written as @xmath12 in this work , we consider only a special case for the condition : @xmath13 .
then , @xmath14 , and @xmath15 .
the solutions @xmath16 and @xmath17 are therefore reduced to a solution of the nls equation by the straight - forward transformation .
hence , by n - fold dt , we can generate new solutions @xmath18}$ ] and @xmath19}$ ] of the manakov system from these _ seed _ solutions , which satisfy the condition @xmath20}_1=
\alpha q^{[n]}_2 $ ] , and then they are automatically reduced to new solutions @xmath20}$ ] of the nls .
now , we solve the eigenfunctions of the lax pair , associated with _ seed
_ solutions .
firstly , we construct the following transformation @xmath21 to map the variable coefficient partial differential equations to constant coefficient partial differential equations @xmath22 by the following matrix @xmath23 here , @xmath24 @xmath25 @xmath26 , @xmath27 denotes the transposition operator . secondly , taking into account the dispersion relation in eqs . , and , then @xmath28 are solved by the method of characteristic equation . here ,
@xmath29 are three constants , @xmath30,@xmath31 and @xmath32 are three roots of the following equation @xmath33 with @xmath34 finally , solving eq . , we have @xmath35 thus , eigenfunctions of the manakov system associated with _ seed _
. are obtained by taking eq . , eq .
and eq . back into @xmath21 .
in this section , we discuss the solutions of the manakov system and the nls equation . based on early work , where rw solutions have been derived @xcite , we set @xmath36 .
note that @xmath37 and @xmath38 if @xmath39 .
furthermore , we set @xmath40 , @xmath41 and @xmath42 , it is trivial to verify that @xmath43 .
thus , the taylor expansion with respect to @xmath44 of the new solutions @xmath18}$ ] and @xmath19}$ ] generates the vector rws of the manakov system . for convenience ,
we introduce @xmath45 , associated with @xmath1 distinct eigenvalues @xmath46 .
we refer to @xcite for the use of the dt to solve the manakov system and we shall not repeat this procedure in this work . as mentioned earlier , we set @xmath40 , @xmath41 , @xmath42 and @xmath47 in @xmath48}$ ] and @xmath49}$ ] , then first - order taylor expansion of them gives the first - order rw of the manakov system .
[ thm_1rw ] let @xmath50 be a solution of the lax pair equations associated with @xmath51 and _ seed _ solution , @xmath52 , then @xmath53},q_2^{[1]})$ ] given by the following determinant representation are the first - order vector rw of the manakov system : @xmath54}=\left(c_1 + 2{\rm i}\frac{|\omega_{1}'|}{|\omega_{2}'|}\right){\rm e}^{{\rm i}(ax+bt)},\quad q_2^{[1]}=\left(c_2 - 2{\rm i}\frac{|\omega_{3}'|}{|\omega_{2}'|}\right){\rm e}^{{\rm i}(ax+bt)},\ ] ] where @xmath55 and @xmath56 substituting into of heorem [ thm_1rw ] , the explicit expressions of the the first - order vector rws of the manakov system are given as @xmath57}=e^{{\rm i}(ax+bt)}(c_1+\frac{f_1}{h_1}),\quad q_2^{[1]}=e^{{\rm i}(ax+bt)}(c_2+\frac{g_1}{h_1 } ) , \end{aligned}\ ] ] where @xmath58 it is obvious that @xmath59}}/{q_2^{[1]}}={c_1}/{c_2}$ ] , so that @xmath60}=\sqrt{1+\frac{c^2_2}{c^2_1}}q_1^{[1]}=\sqrt{1+\frac{c^2_1}{c^2_2}}q_2^{[1]}$ ] is the first - order rw of the nls , which is plotted in fig .
[ fig1 ] . by choosing @xmath61 ,
the amplitude of the plane wave , i.e. carrier of @xmath60}$ ] is of one .
the maximum amplitude of the rw @xmath60}$ ] is located at @xmath62 .
compared with the results of @xcite ( eqs .
10(a ) and 10(b ) , figs . 1 , 2 and 3 )
, it can be concluded that @xmath60}$ ] possesses the same dynamics as the doubly - localized peregrine solution @xcite , except for the location of the rogue peak .
note that @xmath60}$ ] is non - symmetric with respect to ( @xmath63)-axis , as displayed in fig .
however , after applying a straight - forward transformation @xmath64 , it becomes symmetric .
moreover , let @xmath65 and @xmath66 in this case , then @xmath67 , and @xmath68}=\sqrt{2}q_1^{[1]}=\sqrt{2}q_2^{[1]}={\frac { \left ( 3 - 16\,{t}^{2}+16\,{\rm i}t-4 \,{x}^{2 } \right ) } { 1 + 16\,{t}^{2}+4\,{x}^{2}}}{{\rm e}^{2\,{\rm i}t}}\ ] ] is the peregrine soliton of the nls equation @xcite , as expected , see fig .
[ fig2 ] .
this fact means that the asymmetry feature in the first - order rw is non - essential , since it can be explained by a simple linear shift of the spatial variable .
we need to study higher - order rws of the nls to explore the non - symmetric property of such solutions .
similar to the case of nls @xcite and according to the one - fold dt , we obtain the determinant expressions of @xmath1th - order solution @xmath69},\,q_2^{[n]})$ ] through iteration . with the method of taylor expansion with respect to @xmath44 in @xmath69},\,q_2^{[n]})$ ] through @xmath70 ,
the determinant expressions of the @xmath1th - order rws are obtained as well .
[ thm_nrw ] let @xmath71 @xmath72 be @xmath1 distinct eigenfunctions associated with @xmath73 , then , @xmath69},\,q_2^{[n]})$ ] given by the following determinant representations are the @xmath1th - order rws of the manakov system .
@xmath74}=(c_1 + 2{\rm i}\frac{|\omega_{1}'|}{|\omega_{2}'|}){\rm e}^{{\rm i}(ax+bt)},\quad q_2^{[n]}=(c_2 - 2{\rm i}\frac{|\omega_{3}'|}{|\omega_{2}'|}){\rm e}^{{\rm i}(ax+bt)},\ ] ] where @xmath75 @xmath76 and @xmath77 @xmath78 + 2,\,[i]$ ] denotes the floor function of @xmath79 . for convenience ,
let @xmath80 and @xmath66 in the following
. then , @xmath67 , and @xmath20}=\sqrt{2}q_1^{[n]}=\sqrt{2}q_2^{[n]}$ ] is the @xmath1th - order rws of the nls equation .
let @xmath81 in heorem [ thm_nrw ] , the second - order vector rws of the manakov system are given by @xmath82}=q_2^{[2]}=(\frac{\sqrt{2}}{2}+\frac{f_2}{h_2}){\rm e}^{2{\rm i}t } , \end{aligned}\ ] ] with @xmath83 it is a non - symmetrical doubly - localized rw @xmath84}=\sqrt{2}q_1^{[2]}$ ] of the nls , and is displayed in fig .
[ fig3 ] .
it is obvious that the structure of this solution is highly non - symmetrical in the @xmath85 plane and the highest peak is not located at @xmath86 , unlike the profile of the symmetrical doubly - localized ap breathers fundamental pattern @xcite .
in fact , as can been observed qualitatively in fig . [ fig3 ]
( a ) , there are four small soliton peaks surrounding one main and highest peak . however , the left two soliton peaks with respect to the ( @xmath63)-axis are higher than the two on the right .
besides , the highest peak is not located on the center of the four small peaks as the symmetrical fundamental pattern@xcite
. it is located on the left side of the co - ordinate origin center . specifically , the co - ordinate of the main peak is @xmath87 in the @xmath85 plane , and the height of the main peak is of 4.695 .
this value corresponds to the amplitude amplification of this solution , since the carrier amplitude is of one .
this also means that the amplitude of this rw is not distributed symmetrically along @xmath88-axis . in fig .
[ fig3 ] ( a ) , it can be also seen that there is only one significant trough at the left side of the main peak .
this is not the case for the symmetric ap solution , where two identical and significant troughs can be noticed at each side of the main peak along ( @xmath89)-axis , see @xcite .
in order to depict these troughs , we have displayed the contour line at asymptotic plane in fig .
[ fig3 ] ( b ) . generally , fig .
[ fig3 ] ( a ) and ( b ) show the noticeable and clear differences to the dynamics , described by the symmetrical fundamental pattern of the second - order nls rw @xcite .
the centers of the depicted three circles are given by ( 0.6,0.41 ) , ( 0.84 , 0 ) , ( 0.6,-0.41 ) in the ( @xmath90)-plane .
next , let @xmath91 in theorem [ thm_nrw ] , then the third - order rw of manakov system is given as @xmath92}=q_2^{[3]}=e^{2{\rm i}t}(\frac{\sqrt{2}}{2}+\frac{f_3}{h_3}).\ ] ] here , @xmath93 and @xmath94 are two polynomials with degree 12 of @xmath88 and @xmath95 , which are given in appendix .
because @xmath96 when @xmath97 and @xmath98 , the asymptotical height of @xmath99}|$ ] and @xmath100}|$ ] is @xmath101 .
once again @xmath102}=\sqrt{2}q_1^{[3]}=\sqrt{2}q_1^{[3]}$ ] gives the third - order non - symmetrical rw of the nls , which has asymptotical height @xmath103 when @xmath97 and @xmath98 .
the profile and the density plot of @xmath102}$ ] are displayed in fig .
[ fig4 ] , which as expected are different from the symmetrical pattern of the third - order ap @xcite .
similar to the case of @xmath84}$ ] , it is clearly not symmetric with respect to the spatial co - ordinate , however still the symmetry is conserved with respect to the ( @xmath89)-axis . the co - ordinate of main peak of @xmath104}|$ ] is indeed not at @xmath105 .
the maximum amplitude occurs at @xmath106 in the ( @xmath90)-plane , is of @xmath107 .
this value is smaller than 7 , which characterizes the third - order ap solution @xcite .
note also that the latter solution maps six zero amplitude points located on the @xmath88-axis .
since the dt of the manakov system requires solving a cubic algebraic equation , thus , it is possible to derive novel non - trivial solutions for the nls , when the condition described in eq . applies .
indeed , it would be also possible to derive the presented doubly - localized solutions directly from the nls .
this needs further work , which has been started . in the section
, we analyze the differences between the derived solution to the family of aps up to third - order .
since the non - symmetric solution of manakov system can generate the solutions of the nls equation through the simple scaling transformation @xmath20}=\sqrt{2 } q_1^{[n]}$ ] , when @xmath61 in eqs . , we just discuss the difference between symmetrical and non - symmetrical solution of the nls equation .
the main characteristic features are summarized in table 1 . .
the difference between symmetric @xmath108}$ ] and non - symmetric @xmath109}$ ] rws of the nls equation . here , _ no . zeros _ denotes the number of zero amplitudes in @xmath108}\&q^{[j]}$ ] along @xmath88-axis , whereas _ denom .
labels the degree of the polynomials of @xmath88 and @xmath95 in the denominators of @xmath108}\&q^{[j]}$ ] .
furthermore , @xmath108}$ ] and @xmath109}$ ] represent the symmetric and non - symmetric solution , respectively .
the symmetrical fundamental patterns @xmath108}$ ] have a fixed asymptotical plane with a scaled height of @xmath103 , as constructed in @xcite . [ cols="^,^,^,^,^,^",options="header " , ] + as already mentioned , we can conclude that the solutions @xmath109}(j=2,3)$ ] ( the non - symmetric solution ) are just symmetrical with respect to the @xmath95-axis and not symmetrical with respect to the @xmath88-axis .
furthermore , the amplitude of the non - symmetric solution is slightly smaller than the corresponding symmetric ap case .
besides , zero amplitudes in @xmath110}(x,0)(j=2,3)|$ ] are not distributed along @xmath88-axis as in the case of symmetrical rws@xcite .
the number of the zero amplitudes of the rws @xmath20}$ ] is increasing by increasing the order @xmath1 .
therefore , it is reasonable to infer that the non - symmetric feature becomes more and more remarkable and noticeable , thus , such rw solutions deserve further studies .
this section is dedicated to the evolution of the derived second - order non - symmetrical rw @xmath84}$ ] in dimensional physical units and to laboratory experiments , which have been conducted in order to observe the latter nls solution . for that purpose
the solution has to be written first in dimensional units , in order to satisfy the deep - water nls @xcite : @xmath111 while the wave frequency @xmath112 and the wave number @xmath113 are connected through the dispersion relation @xmath114 , where @xmath115 labels the gravitational acceleration .
then , the spatio - temporal evolution of the free surface elevation @xmath116 is approximated to first - order in steepness : @xmath117\right ) .
\label{se}\ ] ] the amplitude of the carrier @xmath118 and its steepness @xmath119 have been chosen to be of 0.003 m and 0.05 , respectively .
therefore , the wave frequency is @xmath120 .
these values have been chosen in order to satisfy the deep - water gravity waves conditions and to avoid wave breaking , which would significantly violate the evolution of the wave field , as predicted and approximated by the nls .
the first step consists in determining the appropriate initial spatial co - ordinate for the boundary conditions eq .
( [ se ] ) , which are needed in order to drive the physical wave maker .
therefore , preliminary numerical simulations based on the modified nls ( mnls ) @xcite , known to describe more accurately the evolution of nonlinear wave packets compared with the nls since it takes into account higher - order dispersion as well as the mean flow of the wave field , have been performed for the dimensional non - symmetrical second - order @xmath121}$ ] .
note that as described earlier and differently than the second - order ap solution the maximal amplification of the corresponding field is not at @xmath122 .
results of the simulations , showing the last stage of envelope propagation over 50 m , before reaching its maximal amplification and for the initial spatial co - ordinate @xmath123 m , are depicted in fig .
[ fig5 ] . }
$ ] for the carrier ampliude @xmath124 m and @xmath125 , while evolving in space with the group velocity , which is @xmath126.,width=377 ] results in fig .
[ fig5 ] show the evolution of the @xmath121}$]-wave field , which initial perturbation consists of two small envelope modulations , as described by nls boundary conditions . during this propagation with the group velocity ,
a strong maximal localization is formed after 50 m and is the result of the nonlinear interaction of the two initially small envelopes .
as next , we will trigger the same evolution dynamics in a laboratory wave flume .
experiments have been conducted in the same facility , as described in detail in @xcite .
the carrier parameters as well as initial conditions have been chosen to be the same as for the numerical simulations of the evolution dynamics .
the wave gauge is placed 10 m from the wave maker , which consists of a single flap , controlled by a hydraulic cylinder .
the from the gauge measured wave profiles are then reemitted to the flap in order to overcome the length restriction of the wave flume facility and to increase its length , as described in @xcite . therefore
, four iteration loops have to be conducted to satisfy a propagation distance of 50 m. the experimental results for the same initial condition as for the numerical simulations are shown in fig . [ fig6 ] . }
$ ] in a water wave tank for @xmath124 m and @xmath125 .
the five measurements are aligned by the value of the group velocity.,width=377 ] as for the doubly - localized second - order ap solution @xcite and as already provided by the mnls simulations , the characteristic amplification of the wave field is due to the overlapping two envelope modulations .
however , the difference with respect to the symmetric ap solution is that this nonlinear envelope interaction is non - symmetric already at the initial stage of the small amplitude modulations , as can be noticed in fig .
this asymmetrical interaction persists for the whole propagation distance along the flume .
indeed , as predicted by the mnls simulations , the maximal amplitude amplification is reached after 50 m of wave propagation .
next , we compare the maximal wave profile with the nls prediction . fig .
[ fig7 ] shows the results .
clearly , the agreement is remarkable .
the observed and theoretical wave profiles , shown in fig .
[ fig7 ] are _ almost _ identical with a noticeable asymmetry compared to the maximal wave , related to this particular solution .
furthermore , the maximal characteristic amplitude amplification of 4.6 is reached .
these experimental observations prove that higher - order nonlinear envelopes , generating non - symmetric rws , can be described by weakly nonlinear evolution equations and can be observed in nonlinear dispersive media , governed by the nls .
the derived non - symmetric solutions naturally have different physical and hydrodynamic properties in terms of surface wave profiles as well as flow field variations , compared with the symmetric ap solutions .
this should be analyzed and investigated in more detail numerically and experimentally in order to quantify these physical differences .
furthermore , the results may motivate similar analytical , numerical and experimental studies in other media , governed by the nls , such as in optics or in plasma .
to summarize , we presented the determinant expressions of the @xmath1th - order rws of the manakov system in theorem 1 and theorem 2 , and gave the explicit expressions for the first- , second- and third - order rw solutions in a particular case such that @xmath127}=q_2^{[j]}(j=1,2,3)$ ] .
as special case , these three solutions determine non - symmetrical rws @xmath109}(j=1,2,3)$ ] with respect to the spatial co - ordinate under the fundamental framework of the nls
. fundamental characteristics of these doubly - localized nls solutions as well as the differences to the corresponding symmetrical ap case , have been discussed in detail .
furthermore , we confirmed the physical validity of the second - order solution @xmath84}$ ] , by performing numerical simulations as well as laboratory experiments .
the theoretical , numerical and experimental results are in a very good agreement and confirm the accuracy of weakly nonlinear evolution equations to describe symmetric and non - symmetric strongly localized pattern dynamics . this work may motivate further numerical simulations of the fully nonlinear hydrodynamic evolution equations and further experiments in different nonlinear dispersive media . *
acknowledgments * this work is supported by the nsf of china under grant no.11271210 , the k. c. wong magna fund in ningbo university and the natural science foundation of ningbo under grant no .
j. s. h acknowledges sincerely prof .
s. fokas for arranging the visit to cambridge university in 2012 - 2014 and for many useful discussions .
a. c. acknowledges partial support from the australian research council ( discovery projects no .
dp1093349 ) and support from the isaac newton institute for mathematical sciences . 00 c. kharif , e. pelinovsky , and a. slunyaev , _ rogue waves in the ocean _
( springer , 2009 ) . c. kharif and e. pelinovsky , eur .
j. mech./ b - fluids 22 , 603 - 634 ( 2003 ) .
a. osborne , _ nonlinear ocean waves and the inverse scattering transform _
( elsevier , 2010 ) .
l. stenflo and p. k. shukla , j. plasma phys .
75 , 841(2009 ) .
d. r. solli , c. ropers , p. koonath , and b. jalali , nature(london)450 , 1054(2007 ) .
b. kibler , j. fatome , c. finot , g. millot , g. genty , b. wetzel , n. akhmediev , f. dias and j. m. dudley , sci .
rep . 2 , 463(2012 ) .
w. m. moslem , p. k. shukla , and b. eliasson , europhys . lett .
96 , 25002(2011 ) .
s.k . sharma and h. bailung , j. geophys .
space phys . 118 919924(2013 ) .
n. akhmediev , a. ankiewicz , and m. taki , phys .
a 373 , 675(2009 ) . v.
e. zakharov and a. a. gelash , phys .
111 , 054101 ( 2013 ) m. onorato , s. residori , u. bortolozzo , a. montinad , f.t .
arecchi , phys .
528 , 4789(2013 ) .
d. h. peregrine , j. aust . math
series b 25 , 16(1983 ) e. a. kuznetsov , sov .
22 , 507508 ( 1977 ) . y. c. ma , stud .
60 , 4358 ( 1979 ) .
n. akhmediev , v. m. eleonskii , and n. kulagin , j. exp .
61 , 894899 ( 1985 ) .
n. akhmediev and v. i. korneev , theor .
phys . 69 , 10891093(1986 ) .
b. kibler , j. fatome , c. finot , g. millot , f. dias , g. genty , n. akhmediev , and j. m. dudley , nat .
phys . 6 , 790(2010 ) .
k. hammani , b. kibler , c. finot , p. morin , j. fatome , j .
m. dudley , and g. millot , opt . lett .
36 , 112114(2011 ) .
a. chabchoub , n. p. hoffmann , and n. akhmrdiev , phys .
106,204502(2011 ) .
a. chabchoub , n. p. hoffmann , and n. akhmediev , phys.rev.e 86 , 016311(2012 ) .
h. bailung , s. k. sharma , and y. nakamura phys .
107 , 255005(2011 ) .
a. chabchoub , n. p. hoffmann , m. onorato , and n. akhmediev , phys .
x 2 , 011015(2012 ) .
a. chabchoub , n. p. hoffmann , m. onorato , a. slunyaev , a. sergeeva , e. pelinovsky , and n. akhmediev , phys .
e 86 , 056601(2012 ) .
a. chabchoub and n. akhmediev , phys .
lett . a 377 , 2590 - 2593(2013 ) .
b. frisquet , b. kibler , and g. millot , phys .
x 3 , 041032 ( 2013 ) .
b. frisquet , a. chabchoub , j. fatome , c. finot , b. kibler , and g. millot , phys .
rev . a 89 , 023821 ( 2014 ) .
d. j. kedziora , a. ankiewicz , and n. akhmediev , phys . rev .
e 88 , 013207 ( 2013 ) .
k. b. dysthe , proc .
r. soc . a 369 , 105114 ( 1979 ) .
k. trulsen and k. b. dysthe wave motion 24 , 281289 ( 1996 ) .
a. chabchoub , n. hoffmann , m. onorato , g. genty , j. m. dudley , and n. akhmediev , phys .
111 , 054104 ( 2013 ) .
y. v. bludov , v. v. konotop , and n. akhmediev , eur .
. j. special topics 185 , 169180(2010 ) .
b. l. guo and l. m. ling , chin .
28 , 110202(2011 ) .
qin , m. gu , phys .
e 86 , 036601(2012 ) .
f. baronio , a. degasperis , m. conforti , and s. wabnitz , phys .
109 , 044102(2012 ) .
l. c. zhao and j. liu , phys .
e 87,013201(2013 ) .
a. degasperis and s. lombardo , phys .
e 88 , 052914(2013 ) .
b. g. zhai , w. g. zhang , x. l. wang , and h. q. zhang , nonlinear anal - real 14 , 14(2013 ) .
s. c. chen and l. y. song , phys .
rev.e 87 , 032910(2013 ) j. s. he , s. w. xu , and k. porsezian , j. phys .
japan 81 , 033002(2012 ) .
j. s. he , s. w. xu , and k. porsezian , phys.rev.e 86 , 066603(2012 ) . c. z. li , j. s. he , and k. porsezian , phys.rev .
e 87 , 012913 ( 2013 ) .
s. w. xu , k. porsezian , j. s. he , and y. cheng , phys .
e 88 , 062925(2013 ) .
s. v. manakov , sov .
jetp 38 , 248 - 253(1974 ) .
m. onorato , a. r. osborne , and m. serio , phys . rev
96 , 014503 ( 2006 ) .
o. gramstad and k. trulsen , phys .
fluids 23 , 062102 ( 2011 ) .
n. akhmediev and a. ankiewicz , solitons : nonlinear pulses and beams ( chapman & hall , london , 1997 )
. g. neugebauer and r. meinel , phys . lett . a 100 , 46770(1984 ) .
y. s. li , x. s. gu , and m. r. zou , acta mathematica sinica ( new series ) 3 , 143151(1985 ) . v. b. matveev and m. a. salle , darboux transformations and solitons ( new york : springer)(1991 ) . c. h. gu , h. s. hu and z. x. zhou , darboux trasformations in integrable systems ( dortrecht : springer)(2006 ) .
he , l. zhang , y. cheng , and y. s. li , sci . china a 12 , 186778(2006 ) .
a. degasperis and s. lombardo , j. phy .
a : mathematical and theoretical 40 , 961977(2007 ) . a. degasperis and s. lombardo , j. phy . a : mathematical and theoretical 42 , 385206(2009 ) .
l. zhang , j. s. he , y. cheng , and y.s .
li , acta mathematica sinica(english series ) 28 , 17131726(2012 ) .
n. akhmediev , a. ankiewicz , and j. m. soto - crespo , phys .
e 80 , 026601(2009 ) .
b. l. guo , l. m. ling , and q. p. liu , phys .
e 85 , 026607(2012 ) .
j. s. he , h. r. zhang , l. h. wang , k. porsezian , and a. s. fokas , phys .
e 87 , 052914 ( 2013 ) .
o. c. wright and m. g. forest , physica d 141 , 104116(2000 ) .
s. w. xu , j. s. he , and l. h. wang , j. phys . a 44 , 305203(2011 ) .
s. w. xu and j. s. he , j. math .
53 , 063507(2012 ) .
y. s. zhang , l. j. guo , s. w. xu , z. w. wu , and j. s. he , commun nonlinear sci numer simulat 19 , 17061722(2014 ) .
l. j. guo , y. s. zhang , s. w. xu , z. w. wu , and j. s. he , phys.script 89 , 035501(2014 )
. n. akhmediev , j.m .
soto - crespo , a .
ankiewicz , phys .
lett . a 373 , 2137(2009 ) .
v. e. zakharov , j. appl .
phys . 9 , 190194 ( 1968 ) . | we present determinant expressions for vector rogue wave solutions of the manakov system , a two - component coupled nonlinear schrdinger equation . as special case
, we generate a family of exact and non - symmetric rogue wave solutions of the nonlinear schrdinger equation up to third - order , localized in both space and time .
the derived non - symmetric doubly - localized second - order solution is generated experimentally in a water wave flume for deep - water conditions .
experimental results , confirming the characteristic non - symmetric pattern of the solution , are in very good agreement with theory as well as with numerical simulations , based on the modified nonlinear schrdinger equation , known to model accurately the dynamics of weakly nonlinear wave packets in deep - water
. corresponding author : [email protected] , [email protected] ] |
in the last decades , more than 850 exoplanets have been detected outside the solar system , while thousands of potential planet candidates from the kepler mission are waiting for confirmation .
even though most of the discovered exoplanets are gas giants , as the larger planets are easier to detect than the smaller rocky ones , evolving observational capabilities have already allowed us to discover tens of planets in the super - earth mass range ( e.g. @xcite ) , some of them probably lying within the habitable zone of their stars ( @xcite ) .
moreover , some earth - sized , and even smaller , exoplanets have already been reported in the literature ( @xcite ) .
indeed , early statistics indicate that about 62% of the milky way stars may host a super - earth ( @xcite ) .
thus , one can confidently expect that true earth analogues will be discovered in large numbers in the near future . to be prepared for the characterization of future exoearth detections , the exploration of our own solar system and its planets is essential .
this will allow us to test our theories and models , enabling more accurate determinations , and characterization of the exoplanets atmospheres and surfaces .
in particular , observation of the solar system rocky planets , including earth , will be key for the search for life elsewhere . over the last years , a variety of studies ,
both observational and theoretical , to determine how the earth would look like to an extrasolar observer have been carried out .
one of the observational approaches has been to observe the earthshine , i.e. , the sunlight reflected by earth via the dark side of the moon .
the visible spectrum of the earthshine has been studied by several authors ( @xcite ) , while more recent studies have extended these observations to the near - infrared ( @xcite ) and to the near - uv ( @xcite ) .
several authors have also attempted to measure the characteristics of the reflected spectrum and the enhancement of earth s reflectance at 700 nm due to the presence of vegetation , known as red - edge , directly ( @xcite ) , and also by using simulations ( @xcite ) .
the red - edge has been proposed as a possible biomarker in earth - like planets ( @xcite ) .
furthermore , @xcite determined that the light scattered by the earth as a function of time contains sufficient information , even with the presence of clouds , to accurately measure earth s rotation period .
more recently , @xcite have also studied the use of the linear polarization content of the earthshine to detect biosignatures , and were able to determine the fraction of clouds , oceans , and even vegetation .
another approach has been to analyze earth s observations from remote - sensing platforms ( e.g. , @xcite ) .
these kinds of observations have also allowed the possibility to reconstruct the continental distribution of our own planet from scattered light curves . @xcite
performed principal components analysis in order to reconstruct surface features from the epoxi data .
@xcite and @xcite proposed an inversion technique which enables to sketch a two - dimensional albedo map from annual variations of the disk - averaged scattered light . in addition , @xcite attempted to reproduce a longitudinal map of the earth from simulated photometric data by using the difference in reflectivity between land and oceans .
other authors have also studied what the color of an extreme earth - like planet might be by utilizing filter photometry ( @xcite ) , while others have studied the disk - averaged spectra of cryptic photosynthesis worlds ( @xcite ) . however , it is unlikely that , even if we were to find an earth - twin , planet will be at an evolutionary stage similar to the earth is today . on the contrary
, extrasolar planets are expected to exhibit a wide range of ages and evolutionary stages .
because of this , it is of interest not only to use our own planet , as it is today , as an exemplar case , but also at different epochs ( @xcite ) .
the development of advanced plants is believed to have taken place on earth during the late ordovician , about 450 ma ago , albeit fungi , algae , and lichens may have greened many land areas before then ( @xcite ) .
microbial mats are multilayered sheets of microorganisms generally composed of both prokaryotes and eukaryotes , being able to reach a thickness of a few centimeters .
the time when microbial mats appeared on the earth surface is still not clear , but prior to the evolution of algae and land plants on early earth , photosynthetic microbial mats probably were among the major forms of life on our planet .
microbial mats are found in the fossil record as early as 3.5 billion years ago .
later , when advanced plants and animals evolved , extensive microbial mats became rarer , but they are still presented in our planet in many ecosystems ( @xcite ) .
even today , they still persist in special environments such as thermal springs , high salinity environments , and sulfur springs .
in this paper , we aimed to discern the effect that the evolution of life over land might have had on the way our planet would look like to a remote extraterrestrial observer . to this end , we have simulated both visible and near - ir disk - integrated spectra of our planet , considering as scenario of such simulations the earth during the late cambrian ( figure [ mapa ] ) , the period during which the development of the first plants is believed to have taken place .
thus , it is clear that as life evolved over land , earth s continental masses must have changed gradually going through different states that go from deserts , to microbial mats , to evolved plants .
with the aim of deriving disk - averaged spectra of the earth 500 ma ago , at any viewing and illumination geometry , and for several surface types covering the continental crust , we have generated a database of one - dimensional synthetic radiance of the earth , i.e. , we have calculated synthetic spectra for a variety of surface and cloud types , and for several viewing and illumination angles . to do that ,
we have used a line - by - line radiative transfer algorithm , based on the disort ( discrete ordinates radiative transfer program for a multi - layered plane - parallel medium ) code ( @xcite ) .
our radiative transfer model ( rtm ) uses profiles of temperature and atmospheric composition , spectral albedos of each surface type , and cloudiness information as input data for the calculations ( see subsequent subsections for a detailed description of inputs ) .
the code considers earth s atmosphere as plane - parallel , and its radiative properties are prescribed on a grid of contiguous spectral bins .
this spectral grid is unevenly spaced , and designed to resolve the rapid variations in the radiative properties near molecular absorption lines .
each spectrum is calculated at very high spectral resolution , with no less than three points per doppler width .
positions , intensities and lineshape parameters for molecular absorption bands are taken from hitran2008 ( @xcite ) .
only a single angle of incidence and 10 angles of reflection can be used for each model run .
finally , each spectrum is degraded to a lower resolution for storage purposes ( @xmath2 ) .
we have run this radiative code using input parameters covering a broad range of viewing and illumination angles , surface and cloud types , atmospheric profiles , and aerosol concentrations , leading to the generation of a database containing about 7000 spectra .
the rtm is essentially an extension of the rtm for transits described in @xcite and @xcite to a viewing geometry for which the light reaching the observer has been reflected at the planet .
the rtm is also capable of modeling the planet s emission of thermal radiation , although that possibility is not explored here .
once the spectral library was generated , we developed a computer code to calculate the disk - integrated irradiance of our planet for given sub - solar and sub - observer points , a surface map , and a cloudiness distribution . at any given date
the code calculates both the fraction of the planet visible to the observer and the illuminated fraction of the earth from the location of the observer .
the planets is subdivided in a 64x32 pixel ( longitude by latitude ) grid .
this grid resolution offers the best balance between computational cost and accuracy . for each grid a radiance spectrum from the aforementioned database is assigned to each detectable and illuminated surface s pixel taking into account the surface type , and the percentage cloud type and amount of such pixel .
note that the full spectral database was generated only for 10 solar angles and for 10 observer angles ( 85@xmath3 , 80@xmath3 , 75@xmath3 , 70@xmath3 , 65@xmath3 , 55@xmath3 , 45@xmath3 , 30@xmath3 , 15@xmath3 , and 0@xmath3 for both angles ) .
thus , radiance values for arbitrary solar and observer angles were interpolated into the angles from the calculated radiances .
finally , to get the disk - averaged spectrum of the ancient earth , we have weighted each pixel s spectrum by their solid angle and we have integrated over the whole earth .
information about temperature and distribution of atmospheric gases were taken from fscatm ( @xcite ) .
we have considered five atmospheric profiles models : tropical , midlatitud summer , midlatitud winter , subartic summer , and subartic winter .
these atmospheric profiles include mixing ratios of the most significant molecules in earth s atmosphere : @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 and @xmath9 .
these properties are prescribed into 33 uneven layers , which go from 0 to 100 km height , being the spacing between layers of 1 km near the bottom of the atmosphere , and 5 km or more above 25 km height .
note that earth s atmosphere 500 ma ago can be considered similar to the present one since , on average , the same atmospheric composition and mean averaged temperature have existed during this period ( @xcite ) .
this is a necessary condition for our simulations to hold as a valid approximation . to carry out our goal of finding out how the appearance of life over land could affect earth s reflectance properties ,
we have considered four different continental land types : water , desert , microbial mats and forest .
figure [ albedos ] shows the wavelength - dependent albedos of each surface type according to the aster spectral library and the usgs digital spectral library .
the continental distribution during the late cambrian has been taken from ron blakey s web site .
surface maps of our planet in that epoch are available online .
the earth geologic information has been regridded into the 64x32 pixel grid used by our model . in our model
, the spatial distribution of clouds was taken from the international satellite cloud climatology project ( isccp ; @xcite ) cloud climatology . with the aim of getting information about the global distribution of clouds over different surface types , and in order to reconstruct the possible cloud distribution of the late cambrian ( 500 ma ago ) ,
we have proceeded in the same way as in @xcite but here , instead of using only the total cloud amount information , we have used information on three separate cloud layers : low ( 1000 - 680 mb ) , mid ( 680 - 440 mb ) and high cloud ( 440 - 30 mb ) data . the optical properties of each cloud type , wavelength - dependent scattering and absorption coefficients , and the asymmetry parameter , were obtained from the optical properties of aerosols and clouds ( opac ) data base ( @xcite ) .
we have considered physical cloud thicknesses of 1 km , and we have assumed that the scattering phase function is described by the henyey - greenstein equation inside clouds , and by the rayleigh scattering function outside them .
we calculated the spatial distribution of these three cloud types according to the surface type lying beneath . to perform this classification , we used real geographical information about the different types of surfaces and vegetation in the present earth , available from the isccp web site .
although the isccp classifies the surfaces as water , rain - forest , deciduous - forest , evergreen - forest , grassland , tundra , shrub - land , desert , and ice , we have only distinguished between four surface types : water , desert , ice , and vegetation , where the latest is defined as the sum of rain - forest , deciduous - forest , evergreen - forest , grassland , tundra , or shrub - land .
we then computed the mean cloudiness at each latitude point , obtaining the empirical relationships between the amount of clouds , surface type , and latitude for low , mid , and high clouds ( see @xcite for details on the method ) .
these relationships allow us to reconstruct the possible cloudiness distribution of the earth 500 ma ago .
it is worth noting that we do not have any empirical information about how clouds behave over extended microbial mat surfaces .
thus in our calculation we have used both the cloudiness information corresponding to deserts and vegetation .
probably the answer is somewhere in between , as microbial mats will increase transpiration and moisture compared to bare land .
however , we will show how the choice of cloud amounts does not substantially affect the results .
the development of advanced land plants is believed to have taken place during the late ordovician , about 450 ma ago , albeit fungi , algae , and lichens may have greened many land areas long before ( @xcite ) .
thus , to determine the impact that such changes could cause in the appearance of the ancient earth s spectrum , we have run our model for four separate scenarios .
first of all , to illustrate the spectrum properties of a planet without life on its surface , we have considered that the continental crust of the earth 500 ma ago was entirely desert .
while this might not be entirely true , it certainly was true at some point earlier in earth s history , with an unknown continental distribution . here
, we chose to keep the late cambrian continental structure in order to isolate the spectral changes due to land scenery changes from those introduced by changing continental distribution ( @xcite ) .
then , we have regarded a case where continents were covered by microbial mats . here
, two separate simulations were run , with the cloud types distribution over land corresponding to desert and to vegetated areas .
we have also considered the case where the continental crust during the late cambrian was entirely covered by evolved plants , like those that dominate the continental surface today .
finally , in section [ sec_photometric ] we have also considered more realistic cases with 50% mixtures between different surface types . in all cases ocean areas
remained the same , and a three - layer cloud distribution over the whole planet is applied .
figure [ 500maespectro ] shows synthetic disk - integrated spectra of the earth 500 ma ago covering the spectral ranges between 0.4 and 2.5 @xmath0@xmath1 , at different times of the day , for each of our four cases where continents are totally covered by i ) deserts , ii ) vegetation , iii ) microbial mats with the cloudiness information corresponding to deserts , and iv ) microbial mats with the clouds corresponding to vegetation . in all cases throughout this paper , the observer and the sun are both located over the planet s equator in such a way that the observer is looking at a quarter - illuminated planet ( phase angle @xmath10 ) .
for an extrasolar planet , the maximum angular separation from the parent star along the orbit occurs at phase @xmath10 , as defined from the observer s position .
thus , this is the more relevant geometry for future exoplanet studies .
one can note that as continents come in and out of the field of view , the global earth s spectra change . at 8:00 ut ,
when the continental presence is at maximum , the changes are more dramatic , in contrast to the spectra at 18:00 ut where differences between the four scenarios can not be appreciated .
that is related with the percentage of land that were illuminated and visible at each time .
these percentages are approximately 24% , 46% , 36% , and 6% for 4:00 , 8:00 , 12:00 , and 18:00 ut , respectively . in order to show such intensity changes in terms of the strength of the vegetation s red - edge
, we have calculated the ratio between the 0.740 - 0.750 @xmath0@xmath1 and the 0.678 - 0.682 @xmath0@xmath1 spectral ranges ( table [ tbl-1 ] ) .
the choice of these spectral regions was made according to @xcite .
as expected , at 18:00 ut ( 6% of land ) this rate is almost the same in our four cases , while at 08:00 ut ( 46% of land ) the maximum is reached in the vegetation case where the red - edge is more pronounced .
for a planet covered by bare dirt , figure [ 500maespectro ] illustrates some diurnal temporal variability , which is small in the visible range , but increases toward redder wavelengths .
@xcite studied the mid - ir emission of the planet in order to derive the rotational period .
they found that the signature of the emission from the sahara desert , among other , can be seen from space and determines a clear maximum in emission .
thus , it is probable that this variability would be even larger in the mid - ir . in the case of a planet with continents covered by microbial mats ,
the spectra do not seem to vary much along the planet rotation , despite the fact of having a large and well differentiated land mass .
the impact of using two different clouds layers in the model ( desert or vegetation ) is almost negligible , being the former just a little bit more variable . on the contrary
the spectrum of a planet covered entirely by plants varies significantly along the day , with the maximum variance occurring in the visible and near - ir .
in particular the variability along the red - edge feature ( at 700 @xmath11 ) is large , and it extends all the way into the near - ir .
observing a detailed reflectance spectrum of an exoplanet would no doubt be a major advantage toward characterizing its atmospheric composition and surface features .
however , given the low signal to noise ratio scenarios that are expected for such observations , spectral information might be difficult to retrieve , especially at high enough temporal resolution to allow sampling of the planet s rotation .
broadband photometric observations are perhaps a more realistic scenario . in order to simulate such photometric observations ,
we have convolved our modeled spectra with standard visible and near - infrared photometric filters , namely , b , v , r , i , z , j , h , and k. figure [ hkcomparison ] shows the b - v versus b - i color - color diagram of test planets fully covered by one specific surface type , i.e. , planets totally covered by water , desert , vegetation , and microbial mats , with and without atmosphere ( blue and red symbols , respectively ) . as can be seen , the addition of an atmosphere in our models moves the position of the planets in the color - color diagram significantly , reducing the color spread of the different surface types along the diagram .
figure [ ubvri ] shows the photometric time - dependent variations in the disk - integrated reflected light along one day .
the rotational light curves are plotted for our four scenarios , and for each photometric filter . for the johnson visible system filters ,
these light - curves show a relatively smooth shape , with the four scenarios being quite similar one to another , and with albedo changes lower than 6% in general .
the only noticeable signature is the enhancement in brightness , about 20% , in the i filter , for the vegetated case ( green line ) between 6 and 12 ut .
this is again related to the vegetation s signature , the red - edge , since the main continental mass is in full view ( and illuminated by the sun ) at this time .
more interesting results are obtained when one moves redward ( bottom part of figure [ ubvri ] ) .
the diurnal light curves of a planet with bare desert surfaces ( black line ) in j , h and k filters show a considerable increase in brightness between 6 and 12 ut of about 20% , 30% and 40% respectively , making immediately obvious the existence of a continent .
the same is found in the vegetation case ( green line ) , being these variability of around 30% , 20% and 15% for j , h and k , respectively .
the light curve for a planet with continents covered with microbial mats with cloudiness distribution corresponding to vegetation ( blue line ) is very mutted , with little variability in either the visible or the near - ir spectral ranges .
in contrast , light curves of the microbial mats case with cloudiness distribution corresponding to desert ( red line ) show a decrease in the albedo of around 10% in the h , and k filters , when the main continental mass is facing the observer . in the j
filter these light curves do not exhibit any variation , making it impossible to even discern the presence of continental masses .
figure [ bvri_zjhk ] shows the amplitude of albedo variability of each of the studied cases as a function of the different photometric filters . as considering continents
completely covered by just only one surface type might not be very realistic , we have also analyzed cases where continents were covered by patchy microbial mats , desert and vegetation .
as an example , in figure [ bvri_zjhk ] we present the results obtained for : i ) continental masses irregularly covered by @xmath12 deserts and @xmath12 microbial mats , and ii ) for continents unevenly covered by @xmath12 vegetation and @xmath12 microbial mats .
the curves in figure [ bvri_zjhk ] show a distinctive shape depending on surface type . when continents are completely covered by deserts ( black line ) the amplitude of albedo change increases rapidly when one moves from z filter redward , while the variability is nearly constant in the visible .
when continents are totally covered by vegetation ( green line ) , that albedo variability increases nearly monotonously from r filter to j filter , where it peaks , and then decreases redward .
in contrast to these two cases , when continents are covered by microbial mats ( blue and red lines ) , the amplitude of albedo change is not strongly dependent on the filter selected and is much more mutted .
moreover , when we considered that the earth 500 ma ago had continents covered by a mixture of microbial mats and deserts / vegetation ( yellow and light - blue lines ) , the shape of these amplitude variability curves resembles the typical shape of the desert / vegetated cases , but with muted variability . at the light of these results
we conclude that it should be a priori possible to determine the type of continental surface on a rocky earth - like planet based on color photometry and given a high enough signal - to - noise ratio .
a planet covered by bare desert areas will have a large diurnal variability in the j , h and k bands . while this is also the case for vegetated continents , in this case an additional bump in the i band
is to be expected , which would be missing in a desert planet .
furthermore , the hypothesis of deserted continents could be further tested / confirmed by observations in the mid - ir photometric bands ( @xcite ) . in the case of a planet with extended microbial mats ,
its nature would be determined by the detection of the presence of continents by using the light curves in b and v bands , and then finding a lower than expected percentage variability in the j , h and k bands , with the reflection peak characteristics of plants and deserts missing .
although not shown in this work , we also studied the same scenarios but with cloud - free atmospheres .
as expected , we find that the omission of clouds decreases the amount of reflected light in comparison to the respective non - clear sky condition
. nevertheless , daily variations in the disk - integrated spectra become more dramatic in the cloud - free case than in the cloudy one .
these cases , however , are highly unrealistic .
in this paper we have used a radiative transfer code to simulate the globally - integrated spectral variability of the earth 500 ma ago , using four possible scenarios regarding the continental surface properties .
our simulations also include realistic distributions of clouds over the whole planet at three altitude layers .
we find that as the continental surface changes from desert ground to microbial mats and to land plants , it produces detectable changes in the globally - averaged earth s reflectance that vary substantially as the earth rotates . by binning the data into standard astronomical photometric band
we see that the variability of each surface type is located in different bands and can induce reflectance changes of up to 40% in periods of hours .
we conclude that using photometric observations of an earth - like planet at different photometric bands , it would be possible to discriminate between bare continental surfaces , large microbial mats extensions , or plant - covered continents .
while in the recent literature the red edge feature of vegetation at visible wavelengths has been proposed as a signature for land plants , observations in near - ir bands can be equally suited for this purpose .
the authors thankfully acknowledge the technical expertise and assistance provided by the spanish supercomputing network ( red espaola de supercomputacin ) , as well as the computer resources used : the la palma supercomputer , located at the instituto de astrofsica de canarias .
the authors also acknowledge support from the spanish miciin , grant cgl2009 - 10641 and aya2010 - 18080 .
crrrrr surface type & 2 ut & 8 ut & 12 ut & 18 ut & + desert & 1.14 & 1.35 & 1.16 & 1.03 + forest & 1.03 & 1.05 & 1.03 & 1.01 + microbial mats ( i ) & 1.05 & 1.12 & 1.05 & 1.02 + microbial mats ( ii ) & 1.04 & 1.10 & 1.05 & 1.02 + | understanding the spectral and photometric variability of the earth and the rest of the solar system planets has become of the utmost importance for the future characterization of rocky exoplanets . as this is not only interesting at present times but also along the planetary evolution , we studied the effect that the evolution of microbial mats and plants over land has had on the way our planet looks from afar .
as life evolved , continental surfaces changed gradually and non - uniformly from deserts through microbial mats to land plants , modifying the reflective properties of the ground and most probably the distribution of moisture and cloudiness . here , we used a radiative transfer model of the earth , together with geological paleo - records of the continental distribution and a reconstructed cloud distribution , to simulate the visible and near - ir radiation reflected by our planet as a function of earth s rotation .
we found that the evolution from deserts to microbial mats and to land plants produces detectable changes in the globally averaged earth s reflectance .
the variability of each surface type is located in different bands and can induce reflectance changes of up to 40% in period of hours . we conclude that using photometric observations of an earth - like planet at different photometric bands , it would be possible to discriminate between different surface types . while recent literature propose the red edge feature of vegetation near 0.7 @xmath0@xmath1 as a signature for land plants , observations in near - ir bands can be equally or even better suited for this purpose . |
the circinus molecular cloud is a region of active star formation with a population of young stellar objects ( ysos ) .
previous studies have estimated the distance of the cloud to be several hundred parsecs @xcite , with ( * ? ? ? *
hereafter b99 ) adopting a distance of 700 pc , with a uncertainty of up to 50% .
early work by @xcite identified two embedded objects , with more recent observations uncovering h@xmath0-emitting sources , herbig - haro objects , and molecular outflows @xcite .
the region was observed at millimeter wavelengths by ( * ? ? ? * hereafter r96 ) and b99 .
r96 identified protostellar candidates at 1300 @xmath1 m , while b99 identified large - scale co outflows .
the circinus region provides an good opportunity to investigate a population of young , coeval stars . in this letter
, we present observations by the _ wide - field infrared survey explorer _
( wise ) @xcite of the young cluster associated with the western circinus cloud .
the observations were taken in the four wise bands ( 3.4 , 4.6 , 12 , and 22 @xmath1 m ) .
the nature of the wise dataset makes it useful for the characterization of star forming regions .
bands 3 and 4 ( 12 and 22 @xmath1 m ) , in particular , are well - suited for the detection of warm circumstellar dust and objects embedded in dense material . the wide field - of - view allows one to determine the yso distribution and the structure of nebulosity over a large spatial scale .
the wise dataset also represents a complement to the _ two micron all - sky survey _
( 2mass ) , allowing one to characterize spectral energy distributions ( seds ) of ysos in seven bands from 1.2 to 22 @xmath1 m .
we summarize the observations and data processing in section [ sec - obs ] , present results in [ sec - results ] , and discuss them in [ sec - disc ] .
the wise spacecraft conducts observations in a continuous scanning mode , using a scan mirror to freeze a single pointing on the detector .
individual images are integrated for 8.8 seconds in all four bands .
the images are processed using the wise science data system , developed at the infrared processing and analysis center , to ingest raw data , produce images , detect sources , and extract photometry .
distortion corrections and band merging are also performed by the automated pipeline @xcite .
observations of the western circinus region were taken 2010 february 22 - 24 .
single images , which have a field - of - view of 47@xmath2 and a plate scale of 2.75@xmath3 per pixel , are coadded to produce images free from cosmic rays and other artifacts and a plate scale of 1.38@xmath3 per pixel .
the automated source extraction recovers sources detected at a specified signal - to - noise ratio which , in the case of these observations , is 5 or greater .
positional matching of wise sources to 2mass point sources is automated .
typical positional offsets between wise and 2mass are 0.1 to 0.2@xmath3 . for some sources found in dense groups , band matching was completed by inspection .
automated extraction recovers 6848 sources in a region centered near ( @xmath4 , @xmath5 , j2000 ) , with dimensions of approximately 90@xmath2 by 45@xmath2 .
this list includes a significant amount of contamination , predominantly foreground objects and spurious extractions on nebulosity .
there are also numerous sources with poor photometry ( i.e. , large photometric errors or upper limit measurements ) .
the list was filtered to ensure photometric quality , excluding objects that have photometric errors greater than 0.2 mag or upper limits in band 4 .
there are 644 sources that meet these criteria . out of these
, the vast majority also have band 3 measurements with photometric errors less than 0.2 mag .
it should be noted that several of the objects with valid 22 @xmath1 m photometry but no measurement at 12 @xmath1 m correspond to very red objects in the image ( discussed in [ sec - veryred ] ) .
sources in the final list correspond predominantly to point sources in the band 4 image , though some sources are still a result of extractions on nebulosity .
these can be filtered by omitting the lowest signal - to - noise sources ; requiring snr @xmath6 12 is effective in eliminating contamination while leaving real sources intact .
this signal - to - noise cut roughly corresponds to a 22 @xmath1 m flux density of 8 mjy .
a few spurious extractions in regions of high confusion ( e.g. , the cir - mms aggregate , see below ) or nebulosity were removed by inspection .
the final source list contains 206 sources which are listed , with their photometry and spectral slope ( see [ sec - class ] ) , in table [ tab - sources ] .
a coadded three - color image is shown in the top - left panel of figure [ fig-3color ] , with a total exposure time of approximately 150 s at the image center .
the image highlights the yso population of the region .
several sources in the region were previously detected by the _ infrared astronomical satellite _ ( iras ) mission , and many of these correspond to individual wise objects , as well as groups of objects .
the iras point sources are identified in figure [ fig-3color ] .
iras sources that correspond to single , uncrowded wise sources are also noted in table [ tab - sources ] .
the ysos in the region are clustered in two main aggregates , around the cir - mms millimeter sources ( r96 ) and source 65b from @xcite , hereafter referred to as vdbh65b .
the northern part of the cluster contains a dense aggregate of ysos , located at @xmath7 , @xmath8 , coincident with a structure uncovered by r96 at 1300 @xmath1 m indicating a core of dense , cold dust , and designated as cir - mms sources 1 through 4 ( we refer to this region as the cir - mms core ) .
iras also shows a point source , co - located with the cir - mms core , 15564 - 6254 .
the iras source is closest to wisepc j150033.91 - 630656.6 , though the large beam of iras would suggest that several of the sources in the aggregate contribute flux to the iras object .
a zoomed wise 12 and 22 @xmath1 m image of the core can be seen in figure [ fig-3color ] .
r96 shows an object with multiple lobes , while the wise image shows 7 point sources detected at 12 and 22 @xmath1 m .
photometry of these objects is unreliable due to source confusion .
the estimated fluxes for the sources are noted in table [ tab - sources ] .
another dense aggregate is located about 45@xmath2 to the southeast of the main aggregate , near @xmath9 , @xmath10 .
this group consists of approximately 5 sources in the 22 @xmath1 m image .
the two brightest objects ( wisepc j150323.80 - 632258.8 and j150328.78 - 632316.4 ) lie near vdbh65b .
this particular region also contains several herbig - haro objects , h@xmath0 emitting stars , and other signs of active star formation @xcite .
this region is discussed further in the following section .
several objects also lie outside the main aggregates , as well as in between them . for 171 objects with wise 22 @xmath1 m and 2mass k@xmath11 photometry , we plot a color - magnitude diagram ( cmd ) in figure [ fig - cmd ] . for each object
, we calculate the infrared spectral slope , @xmath12 , using the 2mass k@xmath11 and wise 22 @xmath1 m data points .
objects are classified into yso classes using the criteria described in @xcite : class i sources have @xmath13 .
flat spectrum sources have @xmath14 .
class ii sources have @xmath15 . the color - magnitude regions which correspond to these classifications
are noted . for objects that do not have k@xmath11 photometry
, we use the wise 3.4 @xmath1 m point instead .
there are 21 class i , 16 flat spectrum , and 35 class ii objects , excluding those objects with unreliable photometry .
the seds confirm the nature of the class i and ii objects , and show the diversity of yso objects in the region .
figure [ fig - sed ] presents examples of seds of ysos in the region , spanning a range of @xmath16
. the distribution of ysos of different classes can be found in figure [ fig - map ] .
class i and flat spectrum objects , as well as the very red objects , appear to be concentrated in a filament beginning roughly nnw of the cir - mms group , continuing to the sse , with the cir - mms region at its center .
this line turns towards the east in the region between the cir - mms and vdbh65b aggregates , following a nw to se line , continuing se of vdbh65a .
class ii objects appear to be more scattered , with many lying to the east of cir - mms and to the north of vdbh65b .
observations by b99 find several large scale co outflows powered by ysos in the region .
one of the largest and most energetic of the outflows , designated ` flow a ' by b99 , originates from the cir - mms aggregate ( figure [ fig-3color ] ) , slightly west of iras 14564 - 6254 .
the nearest bright source to the iras object is wisepc j150033.91 - 630656.6 , though several bright sources lie in this compact aggregate .
this group of sources is also co - located with the cir - mms core detected at 1300 @xmath1 m ( r96 ) .
the wise observations of the core are consistent with the hypothesis of b99 and r96 , that the heating and driving of the outflow are a result of several sources , rather than a single yso .
this group of ysos represents the densest collection of luminous mid - infrared sources in the region , with a combined flux density for the region near 10 jy in band 4 .
the objects in the aggregate are predominantly class i or flat spectrum objects .
flows ` b ' and ` c ' , originate from iras 14563 - 6301 ( b99 ) . at the location of this source ,
we find two bright mid - infrared sources close to one another .
they are sources wisepc j150022.72 - 631325.4 and j150024.26 - 631337.4 , which are separated by about 14@xmath3 ( 9800 au at 700 pc ) .
j150024.26 - 631337.4 is closest to the position of the iras source , about 5@xmath3 away .
the two wise sources have 22 @xmath1 m fluxes of over 2 jy and nearly 1 jy , respectively . iras 14563 - 6301
has a flux of 4.1 jy at 25 @xmath1 m ; the discrepancy with wise photometry can be explained by the larger beam and bandpass of iras .
b99 hypothesize that the source of the b and c outflows may be a binary , and this is consistent with our finding of two closely separated mid - infrared sources at the origin .
several less energetic flows , designated d through h have been identified by b99 . for flows that originate from an uncrowded region ,
we note the source in table [ tab - sources ] . among the earliest observations of the region were those of @xcite which uncovered two stars embedded in nebulosity at optical wavelengths .
designated 65a and 65b , the latter is associated with an aggregate of ysos described above , lying near source j150323.80 - 632258.8 in this study .
source 65a corresponds to this study s j150058.58 - 631654.9 .
several candidate ysos in the western circinus cloud have been imaged at 10 @xmath1 m by @xcite .
one of these sources is vdbh65a , and the other four sources are part of the aggregate near the cir - mms core .
a number of other h@xmath0 emission line stars were uncovered by @xcite .
fourteen of the objects they identify are in the region covered by this study , and all have significant mid - infrared excess .
the h@xmath0 emitting stars are distributed throughout the region , but half of the objects are found near the dense cir - mms and vdbh65b aggregates .
a few of these stars are also the source of the large scale co outflows discussed in the previous section .
mid - infrared sources that correspond to h@xmath0 emitting stars are noted in table [ tab - sources ] .
the region also contains several herbig - haro objects , specifically hh 76 , 77 , and 139 - 143 @xcite .
several of these sources lie in the vicinity of vdbh65b , as described in @xcite .
however , strong mid - infrared tracers of outflows such as feii lines at 5.3 , 18 , 26 and 35 @xmath1 m and the h@xmath17 rotational line at 17 @xmath1 m @xcite , do not lie within the four wise bandpasses .
the wise images do not show any clear correlation between detected sources and the known hh objects .
seven objects are detected as relatively bright sources in band 4 , but have no matching 2mass sources , and only upper limit measurements in band 3 ( for all but one object , j145959.01 - 625936.7 ) .
these objects appear as the reddest sources in the 3-color image ( figure [ fig-3color ] ) .
those with band 1 photometry have [ 3.4]-[22 ] @xmath18 , and all have [ 4.6]-[22 ] @xmath19 .
these sources are noted in table [ tab - sources ] , along with their estimated [ 4.6]-[22 ] colors .
the list includes one object , j145941.04 - 625757.6 , which was not extracted by the automated pipeline , but was identified by inspection of the image .
these sources are classified as class i objects based upon @xmath13 ; however , at least one of them ( j150238.02 - 631900.3 ) may be a class 0 source , based upon its association with submillimeter emission .
wisepc j150238.02 - 631900.3 corresponds to the location of a dense core described in b99 and may give us insight into the physical nature of these objects .
b99 notes a prominent and visually very opaque core which does not harbor a known ir source or [ @xmath20co outflow ] " which they hypothesize to be a pre - collapse core .
the region appears as a dark ` knot ' in the nebulosity in the wise image at 12 @xmath1 m , but at 22 @xmath1 m a source with a flux of about 30 mjy is detected .
the wise data suggest that this source is possibly in the early stages of evolution , or it could be very deeply embedded , based on its extremely red color .
the reddest object with [ 4.6]-[22 ] = 8.5 , is j150348.39 - 632632.3 , which is located south of the vdbh65b group .
these objects are shown in the bottom center and right panels of figure [ fig-3color ] .
the observational signature of six of the objects is very similar to that of j150238.02 - 631900.3 ( i.e. , bright 22 @xmath1 m detection , with the 12 @xmath1 m image showing a dark knot ) , which leads us to believe that they may also be physically similar .
j145959.01 - 625936.7 has a faint 12 @xmath1 m detection co - located with the 22 @xmath1 m source , and is also located in a dark filamentary structure .
we calculate the infrared luminosity ( l@xmath21 ) for each yso ( 48 class i , flat spectrum , and class ii objects ) , integrated over the wavelength range 1 to 26 @xmath1 m , using the wise and 2mass seds , and assuming a distance of 700 pc .
values of l@xmath21 are shown in table [ tab - sources ] and span the range 0.08 to 30 l@xmath22 .
a distribution of infrared luminosities is shown in the bottom panel of figure [ fig - cmd ] .
the lowest infrared luminosity objects probed by this study are 1 to 2 orders of magnitude more luminous than the lowest - luminosity spitzer sources in recent studies of nearby regions , such as serpens , taurus , and the cepheus flare @xcite .
furthermore , the infrared luminosity distribution of wise detected sources appears to peak at log(l@xmath21/l@xmath22 ) between -0.5 and 0.0 , about an order of magnitude more luminous than those studies .
thus , the typical yso observed in this study is significantly more luminous than those observed in the nearby regions .
these effects are due to lower wise sensitivity compared to spitzer , and greater distance to the circinus region . on the bright end of the luminosity distribution
, we find similar numbers of infrared - luminous objects , with the number of yso falling off sharply where log(l@xmath21/l@xmath22 ) @xmath23 0 , with no sources more luminous than log(l@xmath21/l@xmath22 ) = 1.5 .
the spatial distribution and clustering of ysos can also be compared qualitatively to serpens .
the physical size of serpens is similar to the western circinus region .
@xcite find that class ii and iii objects dominate the population outside of the core , while class i and flat spectrum objects are common in the core .
we find this also to be the case in the circinus region , where class i objects dominate the cir - mms and vdbh65b aggregates , and are more confined spatially to dark nebular filaments .
class ii objects are more evenly distributed throughout the region .
we have presented wise observations of the western circinus molecular cloud core , covering over 1.1 deg@xmath24 in the four wise bands .
these observations have uncovered a population of ysos with the following characteristics : - the population contains class i , ii , iii and flat spectrum objects .
- the dense aggregate in the northern region of the cluster , referred to here as the cir - mms aggregate , corresponds to a previously discovered dense , cold core at 1300 @xmath1 m ( r96 ) .
this aggregate also drives a strong co outflow discovered by b99 .
- we identify several mid - infrared sources which correspond to iras and h@xmath0 sources .
some iras sources have been resolved into multiple member groups .
- we identify several very red objects with bright detections in band 4 ( 22 @xmath1 m ) .
one of these sources corresponds to a dense core of material identified by b99 , and could be a class 0 object . - the objects observed by wise in the circinus region are typically more luminous than those characterized by spitzer in nearby regions such as serpens and taurus .
clustering of yso classes is similar to serpens , where class i objects are found preferentially in dense cores .
this publication makes use of data products from the wide - field infrared survey explorer , a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration .
w.m.l . acknowledges support from wise and the wise science data center .
this research made use of the simbad database , operated at cds , strasbourg , france .
cccccccccccc j145716.77 - 630302.0 & @xmath25 & @xmath26 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 & -1.82 & iii or ms & - & + j145726.27 - 632202.5 & @xmath32 & @xmath33 & @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 & -2.8 & iii or ms & - & + j145726.80 - 633335.6 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & -2.75 & iii or ms & - & + j145733.81 - 630252.3 & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51 & @xmath52 & -2.7 & iii or ms & - & + j145739.30 - 631851.7 & @xmath53 & @xmath54 & @xmath55 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & -2.67 & iii or ms & - & + j145740.79 - 633533.3 & @xmath60 & @xmath61 & @xmath62 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & -2.4 & iii or ms & - & + j145753.68 - 625737.3 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 & -2.75 & iii or ms & - & + j145753.92 - 631233.9 & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & @xmath80 & -1.86 & iii or ms & - & + j145757.66 - 633055.5 & @xmath81 & @xmath82 & @xmath83 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & -2.75 & iii or ms & - & + j145758.89 - 631715.6 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath38 & -2.92 & iii or ms & - & + j145805.27 - 631709.0 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 & -2.72 & iii or ms & - & + j145810.30 - 625201.6 & @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & -2.21 & iii or ms & - & + j145812.83 - 631236.3 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & -2.79 & iii or ms & - & + j145815.57 - 630219.2 & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & @xmath121 & -2.59 & iii or ms & - & + j145827.41 - 625203.0 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & @xmath128 & -2.72 & iii or ms & - & + j145831.10 - 625449.8 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135 & -2.16 & iii or ms & - & + j145832.83 - 630408.8 & @xmath136 & @xmath137 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 & -1.9 & iii or ms & - & iras 14544 - 6252 + j145833.36 - 631401.0 & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & -0.36 & ii & -0.38 & + j145835.09 - 631256.5 & - & @xmath150 & @xmath151 & @xmath152 & @xmath153 & @xmath154 & @xmath155 & 0.84 & i & - & + j145835.68 - 631430.8 & @xmath156 & @xmath157 & @xmath158 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & -0.44 & ii & -0.71 & + j145837.86 - 625337.2 & @xmath158 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & -0.86 & ii & -0.47 & + j145841.36 - 631048.9 & @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 & @xmath174 & -2.51 & iii or ms & - & + j145841.86 - 631207.5 & - & - & - & @xmath175 & @xmath176 & @xmath145 & @xmath177 & 0.15 & flat & - & + j145843.48 - 625341.5 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & @xmath184 & -1.01 & ii & 0.71 & mo94 - 1 + j145848.44 - 630657.1 & @xmath185 & @xmath186 & @xmath187 & @xmath188 & @xmath189 & @xmath190 & @xmath191 & -2.49 & iii or ms & - & + j145849.77 - 631401.0 & @xmath192 & @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 & -1.21 & ii & 1.23 & iras 14547 - 6302 + j145856.01 - 630822.4 & @xmath199 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & -2.12 & iii or ms & - & + j145901.82 - 633252.6 & @xmath206 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & @xmath212 & -1.99 & iii or ms & - & + j145903.59 - 632507.4 & @xmath213 & @xmath214 & @xmath215 & @xmath216 & @xmath217 & @xmath218 & @xmath219 & -2.33 & iii or ms & - & iras 14549 - 6313 + j145910.87 - 625923.3 & @xmath220 & @xmath221 & @xmath222 & @xmath223 & @xmath224 & @xmath225 & @xmath226 & -0.16 & flat & -0.89 & + j145913.60 - 630005.5 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 & 0.64 & i & 0.31 & iras 14551 - 6248 ; mo94 - 2 + j145919.46 - 625151.8 & @xmath234 & @xmath151 & @xmath235 & @xmath236 & @xmath237 & @xmath156 & @xmath238 & -0.5 & ii & -0.95 & + j145919.65 - 633813.7 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & @xmath245 & -0.86 & ii & 1.47 & iras 14551 - 6326 + j145922.50 - 632710.9 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 & @xmath251 & @xmath252 & -2.77 & iii or ms & - & + j145923.06 - 630322.6 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath258 & @xmath259 & -1.82 & iii or ms & - & + j145924.41 - 630832.5 & @xmath260 & @xmath261 & @xmath262 & @xmath263 & @xmath264 & @xmath265 & @xmath266 & -1.9 & iii or ms & - & + j145925.01 - 633906.9 & @xmath267 & @xmath268 & @xmath269 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & -1.98 & iii or ms & - & + j145925.08 - 630018.9 & @xmath274 & @xmath275 & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath280 & -1.63 & iii or ms & - & + j145925.39 - 625121.6 & @xmath187 & @xmath281 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & -2.36 & iii or ms & - & + j145925.58 - 633714.1 & @xmath287 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & -2.78 & iii or ms & - & + j145926.13 - 633608.4 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & @xmath299 & @xmath300 & -1.99 & iii or ms & - & + j145936.09 - 630828.9 & @xmath301 & @xmath302 & @xmath303 & @xmath304 & @xmath305 & @xmath306 & @xmath307 & -2.67 & iii or ms & - & + j145936.85 - 625439.9 & @xmath308 & @xmath309 & @xmath310 & @xmath311 & @xmath312 & @xmath313 & @xmath314 & -2.68 & iii or ms & - & + j145940.08 - 625730.0 & @xmath315 & - & - & @xmath316 & @xmath317 & 0.25u & @xmath318 & 1.71 & i & - & very red ; [ 4.6]-[22]=5.57 + j145940.73 - 632055.8 & @xmath319 & @xmath320 & @xmath321 & @xmath322 & @xmath323 & @xmath324 & @xmath325 & -2.19 & iii or ms & - & iras 14555 - 6308 + j145940.94 - 625955.0 & @xmath326 & @xmath220 & @xmath327 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & -0.26 & flat & -1.09 & + j145941.04 - 625757.6 & - & - & - & - & - & - & - & - & n / a & - & very red ; no photometry + j145944.57 - 631131.8 & @xmath332 & @xmath333 & @xmath334 & @xmath335 & @xmath336 & @xmath337 & @xmath338 & -1.89 & iii or ms & - & iras 14556 - 6259 + j145945.97 - 631434.4 & @xmath339 & @xmath340 & @xmath341 & @xmath342 & @xmath343 & @xmath344 & @xmath345 & -1.77 & iii or ms & - & iras 14556 - 6302 + j145949.82 - 632242.8 & @xmath346 & @xmath347 & @xmath348 & @xmath349 & @xmath350 & @xmath351 & @xmath352 & -2.75 & iii or ms & - & + j145950.07 - 625833.1 & - & - & - & @xmath353 & @xmath354 & @xmath355 & @xmath356 & 1.63 & i & - & + j145951.15 - 625851.3 & @xmath357 & @xmath358 & @xmath359 & @xmath360 & @xmath361 & @xmath362 & @xmath363 & -2.7 & iii or ms & - & + j145955.31 - 630538.5 & @xmath364 & @xmath365 & @xmath366 & @xmath367 & @xmath368 & @xmath369 & @xmath370 & -2.93 & iii or ms & - & + j145959.01 - 625936.7 & - & - & - & @xmath371 & @xmath372 & 0.32u & @xmath373 & 1.33 & i & - & very red ; [ 4.6]-[22]=6.38 + j145959.30 - 631309.7 & @xmath374 & @xmath375 & @xmath376 & @xmath377 & @xmath378 & @xmath379 & @xmath380 & -2.12 & iii or ms & - & + j150000.30 - 632011.9 & @xmath381 & @xmath382 & @xmath383 & @xmath384 & @xmath385 & @xmath386 & @xmath387 & -2.02 & iii or ms & - & + j150000.78 - 631853.6 & @xmath388 & @xmath389 & @xmath390 & @xmath391 & @xmath392 & @xmath393 & @xmath38 & -2.61 & iii or ms & - & + j150002.74 - 632201.2 & @xmath394 & @xmath395 & @xmath396 & @xmath397 & @xmath398 & @xmath399 & @xmath400 & -1.91 & iii or ms & - & + j150003.89 - 630004.9 & @xmath401 & @xmath402 & @xmath403 & @xmath404 & @xmath405 & @xmath406 & @xmath407 & -2.04 & iii or ms & - & + j150004.33 - 625533.0 & @xmath408 & @xmath409 & @xmath410 & @xmath411 & @xmath412 & @xmath413 & @xmath414 & -2.37 & iii or ms & - & + j150009.75 - 633242.2 & @xmath415 & @xmath416 & @xmath417 & @xmath418 & @xmath419 & @xmath369 & @xmath45 & -2.78 & iii or ms & - & + j150015.19 - 632125.9 & - & - & @xmath420 & @xmath421 & @xmath422 & @xmath423 & @xmath424 & 0.16 & flat & - & + j150017.88 - 633544.9 & @xmath425 & @xmath426 & @xmath427 & @xmath428 & @xmath429 & @xmath430 & @xmath431 & -2.08 & iii or ms & - & + j150018.73 - 625959.7 & - & - & - & @xmath421 & @xmath432 & @xmath433 & @xmath434 & 2.95 & i & - & iras
14562 - 6248 + j150021.82 - 630220.3 & @xmath435 & @xmath436 & @xmath437 & @xmath438 & @xmath439 & @xmath440 & @xmath441 & -1.9 & iii or ms & - & + j150022.02 - 631241.9 & @xmath326 & @xmath442 & @xmath51 & @xmath443 & @xmath444 & @xmath445 & @xmath446 & 0.13 & flat & 0.46 & + j150022.72 - 631325.4 & @xmath447 & @xmath448 & @xmath449 & @xmath450221 & @xmath450324 & @xmath450912 & @xmath4502425 & @xmath450 - 0.2 & flat ? & - & c + j150024.04 - 630155.5 & - & - & - & @xmath451 & @xmath452 & @xmath453 & @xmath454 & 0.72 & i & - & iras
14563 - 6250 + j150024.26 - 631337.4 & - & - & - & @xmath45018 & @xmath45031 & @xmath45058 & @xmath450969 & @xmath4501.1 & i ? & - & c + j150024.55 - 631008.2 & @xmath455 & @xmath456 & @xmath457 & @xmath458 & @xmath459 & @xmath460 & @xmath461 & 0.37 & i & 0.44 & c@xmath2 + j150025.45 - 632945.0 & @xmath462 & @xmath463 & @xmath464 & @xmath465 & @xmath466 & @xmath467 & @xmath468 & -2.56 & iii or ms & - & + j150027.85 - 631159.1 & @xmath469 & @xmath470 & @xmath471 & @xmath144 & @xmath401 & @xmath472 & @xmath473 & -0.86 & ii & -0.45 & + j150027.96 - 633620.5 & @xmath474 & @xmath475 & @xmath476 & @xmath477 & @xmath478 & @xmath479 & @xmath480 & -2.38 & iii or ms & - & + j150029.34 - 630945.5 & @xmath481 & @xmath482 & @xmath483 & @xmath484 & @xmath485 & @xmath486 & @xmath487 & -0.72 & ii & 0.23 & mo94 - 4 + j150029.44 - 630742.3 & 0.12u & @xmath488 & @xmath489 & @xmath450134 & @xmath450545 & @xmath450913 & @xmath4503087 & @xmath4500.1 & flat ? & - & c ; cir - mms + j150029.46 - 630629.4 & @xmath315 & - & - & @xmath4504 & @xmath4507 & @xmath45019 & @xmath450100 & @xmath4500.7 & i ? & - & c ;
cir - mms + j150029.84 - 630838.5 & @xmath315 & - & @xmath490 & @xmath491 & @xmath492 & @xmath493 & @xmath494 & 0.22 & flat & - & + j150030.26 - 631418.4 & @xmath495 & @xmath496 & @xmath497 & @xmath498 & @xmath499 & @xmath500 & @xmath501 & -2.04 & iii or ms & - & + j150030.42 - 631017.0 & @xmath502 & @xmath503 & @xmath504 & @xmath505 & @xmath506 & @xmath507 & @xmath508 & -1.35 & ii & -0.05 & + j150030.57 - 631030.3 & @xmath509 & @xmath493 & @xmath510 & @xmath511 & @xmath512 & @xmath513 & @xmath514 & -0.63 & ii & -0.24 & + j150030.78 - 630652.2 & @xmath515 & @xmath516 & @xmath517 & @xmath45076 & @xmath45089 & @xmath450188 & @xmath450450 & @xmath450 - 0.2 & flat ? & - & c ;
cir - mms + j150031.50 - 630703.3 & - & - & - & @xmath45021 & @xmath45055 & @xmath450175 & @xmath450895 & @xmath4501.0 & i ? & - & c ; cir - mms + j150033.91 - 630656.6 & 0.51u & @xmath518 & @xmath519 & @xmath45037 & @xmath450221 & @xmath4501174 & @xmath4505044 & @xmath4501.1 & i ? & - & c ; cir - mms + j150034.43 - 631151.9 & - & - & - & @xmath520 & @xmath521 & @xmath522 & @xmath523 & 1.54 & i & - & + j150034.91 - 630804.4 & - & - & - & @xmath524 & @xmath525 & @xmath526 & @xmath527 & 1.3 & i & - & + j150036.06 - 631315.0 & @xmath528 & @xmath529 & @xmath530 & @xmath531 & @xmath532 & @xmath533 & @xmath534 & 0.82 & i & -0.01 & + j150036.15 - 630512.4 & @xmath535 & @xmath536 & @xmath277 & @xmath537 & @xmath538 & @xmath539 & @xmath540 & -1.19 & ii & 0.07 & + j150037.15 - 630652.2 & @xmath528 & @xmath541 & @xmath542 & @xmath45051 & @xmath450112 & @xmath450182 & @xmath450415 & @xmath4502.1 & i ? & - & c ; cir - mms + j150037.28 - 630718.2 & 0.37u & 0.77u & @xmath543 & @xmath45015 & @xmath45085 & @xmath450126 & @xmath450570 & @xmath4502.4 & i ? & - & c ;
cir - mms + j150037.54 - 630903.6 & - & @xmath544 & @xmath545 & @xmath546 & @xmath547 & @xmath548 & @xmath549 & 0.63 & i & - & + j150038.34 - 631108.2 & - & - & - & @xmath550 & @xmath551 & @xmath552 & @xmath553 & 1.68 & i & - & + j150038.74 - 630743.3 & - & - & @xmath554 & @xmath45012 & @xmath45037 & @xmath4508 & @xmath45030 & @xmath4500.0 & flat & - & c ; cir - mms + j150039.38 - 631231.5 & - & @xmath555 & @xmath556 & @xmath557 & @xmath558 & @xmath559 & @xmath560 & 0.61 & i & - & + j150040.62 - 631522.6 & @xmath561 & @xmath562 & @xmath563 & @xmath564 & @xmath565 & @xmath566 & @xmath45 & -2 & iii or ms & - & + j150041.0 - 630652.9 & 0.18u & @xmath567 & @xmath568 & @xmath45016 & @xmath45034 & @xmath45050 & @xmath450109 & @xmath4500.1 & flat ?
& - & c ; cir - mms + j150041.05 - 630638.9 & 0.08u & @xmath421 & @xmath327 & @xmath4507 & @xmath45087 & @xmath450707 & @xmath4504695 & @xmath4501.2 & i ? & - & c ; cir - mms + j150041.29 - 631138.9 & - & - & - & @xmath488 & @xmath569 & @xmath570 & @xmath571 & 1.73 & i & - & + j150041.66 - 631110.3 & @xmath572 & @xmath573 & @xmath574 & @xmath575 & @xmath576 & @xmath577 & @xmath578 & -0.58 & ii & 0.63 & + j150042.94 - 631030.1 & @xmath317 & @xmath159 & @xmath546 & @xmath579 & @xmath580 & @xmath164 & @xmath581 & -0.08 & flat & -0.65 & + j150043.60 - 631731.9 & @xmath582 & @xmath583 & @xmath584 & @xmath585 & @xmath586 & @xmath587 & @xmath588 & -1.91 & iii or ms & - & + j150043.83 - 630618.4 & @xmath156 & @xmath589 & @xmath590 & @xmath45012 & @xmath45012 & @xmath45013 & @xmath45047 & @xmath450 - 0.4 & ii ? & - & c ; cir - mms1 + j150044.06 - 630106.9 & @xmath591 & @xmath592 & @xmath593 & @xmath594 & @xmath595 & @xmath596 & @xmath597 & -1.99 & iii or ms & - & + j150044.08 - 631349.7 & - & - & - & @xmath598 & @xmath599 & 0.25u & @xmath600 & 1.61 & i & - & very red ; [ 4.6]-[22]=5.50 + j150044.61 - 633851.4 & @xmath601 & @xmath602 & @xmath603 & @xmath604 & @xmath605 & @xmath606 & @xmath607 & -2.69 & iii or ms & - & + j150046.01 - 625427.2 & @xmath608 & @xmath609 & @xmath610 & @xmath611 & @xmath612 & @xmath613 & @xmath614 & -2.84 & iii or ms & - & + j150047.90 - 630614.3 & @xmath615 & @xmath616 & @xmath617 & @xmath618 & @xmath619 & @xmath620 & @xmath621 & -0.97 & ii & -0.34 & + j150048.20 - 631023.1 & @xmath326 & @xmath622 & @xmath623 & @xmath624 & @xmath625 & @xmath626 & @xmath627 & -0.82 & ii & -0.77 & + j150048.68 - 630537.3 & @xmath628 & @xmath629 & @xmath630 & @xmath631 & @xmath632 & @xmath633 & @xmath634 & -0.73 & ii & 0.62 & mo94 - 7 + j150048.95 - 630647.4 & @xmath157 & @xmath531 & @xmath635 & @xmath636 & @xmath637 & @xmath638 & @xmath639 & -0.44 & ii & -0.26 & mo94 - 8 + j150052.69 - 630438.8 & - & - & - & @xmath488 & @xmath640 & @xmath157 & @xmath641 & 0.27 & flat & - & + j150053.83 - 631740.4 & - & - & - & 0.11u & @xmath642 & 0.27u & @xmath643 & - & n / a & - & very red ; [ 4.6]-[22]=7.87 + j150054.73 - 630807.9 & @xmath644 & @xmath645 & @xmath646 & @xmath647 & @xmath648 & @xmath649 & @xmath650 & -0.81 & ii & -0.44 & + j150057.48 - 633144.0 & @xmath651 & @xmath652 & @xmath653 & @xmath654 & @xmath655 & @xmath656 & @xmath657 & -2.64 & iii or ms & - & + j150058.58 - 631654.9 & @xmath658 & @xmath659 & @xmath660 & @xmath661 & @xmath662 & @xmath663 & @xmath664 & 0.02 & flat & 1.11 & iras 14568 - 6304 ; vdbh65a ; mo94 - 10 ; flow f + j150100.11 - 625427.0 & @xmath665 & @xmath666 & @xmath667 & @xmath668 & @xmath669 & @xmath670 & @xmath671 & -2.62 & iii or ms & - & + j150101.54 - 630813.0 & @xmath580 & @xmath672 & @xmath673 & @xmath674 & @xmath675 & @xmath676 & @xmath370 & -1.11 & ii & -0.33 & + j150104.54 - 625557.2 & @xmath677 & @xmath678 & @xmath679 & @xmath680 & @xmath681 & @xmath682 & @xmath683 & -2.82 & iii or ms & - & + j150107.38 - 630335.4 & - & @xmath555 & @xmath529 & @xmath684 & @xmath685 & @xmath686 & @xmath687 & 0.44 & i & - & + j150108.29 - 630350.5 & @xmath688 & @xmath689 & @xmath690 & @xmath691 & @xmath692 & @xmath693 & @xmath694 & -1.19 & ii & -0.21 & + j150112.08 - 633014.6 & @xmath695 & @xmath696 & @xmath697 & @xmath698 & @xmath699 & @xmath700 & @xmath701 & -2.8 & iii or ms & - & + j150114.50 - 630551.1 & @xmath702 & @xmath703 & @xmath704 & @xmath705 & @xmath706 & @xmath707 & @xmath708 & -0.87 & ii & -0.19 & + j150115.48 - 631202.5 & @xmath709 & @xmath710 & @xmath711 & @xmath712 & @xmath713 & @xmath714 & @xmath100 & -2.07 & iii or ms & - & + j150117.08 - 625646.6 & @xmath715 & @xmath716 & @xmath717 & @xmath562 & @xmath718 & @xmath719 & @xmath720 & -1.83 & iii or ms & - & + j150119.42 - 630223.3 & @xmath721 & @xmath722 & @xmath723 & @xmath724 & @xmath725 & @xmath726 & @xmath727 & -1.94 & iii or ms & - & + j150122.83 - 633850.5 & @xmath728 & @xmath729 & @xmath730 & @xmath731 & @xmath732 & @xmath733 & @xmath734 & -2.79 & iii or ms & - & + j150124.66 - 631608.7 & @xmath735 & @xmath736 & @xmath737 & @xmath738 & @xmath739 & @xmath740 & @xmath741 & -2.69 & iii or ms & - & + j150126.44 - 630707.1 & @xmath742 & @xmath743 & @xmath744 & @xmath745 & @xmath746 & @xmath393 & @xmath747 & -2.14 & iii or ms & - & + j150130.91 - 631901.4 & @xmath748 & @xmath749 & @xmath750 & @xmath751 & @xmath752 & @xmath753 & @xmath754 & -2.16 & iii or ms & - & + j150140.06 - 630253.7 & @xmath755 & @xmath756 & @xmath757 & @xmath758 & @xmath759 & @xmath760 & @xmath761 & -2.41 & iii or ms & - & + j150144.37 - 631429.8 & @xmath762 & @xmath763 & @xmath764 & @xmath765 & @xmath766 & @xmath767 & @xmath768 & -2.73 & iii or ms & - & + j150147.36 - 631620.2 & @xmath220 & @xmath769 & @xmath770 & @xmath771 & @xmath772 & @xmath773 & @xmath774 & 0.05 & flat & -0.48 & + j150148.30 - 630343.8 & @xmath775 & @xmath776 & @xmath777 & @xmath778 & @xmath779 & @xmath780 & @xmath781 & -2.67 & iii or ms & - & + j150148.35 - 630310.1 & @xmath782 & @xmath783 & @xmath784 & @xmath785 & @xmath786 & @xmath787 & @xmath788 & -2.14 & iii or ms & - & + j150149.93 - 630448.2 & @xmath789 & @xmath790 & @xmath791 & @xmath792 & @xmath793 & @xmath794 & @xmath795 & -2.63 & iii or ms & - & + j150153.64 - 630611.7 & @xmath796 & @xmath797 & @xmath798 & @xmath799 & @xmath396 & @xmath800 & @xmath801 & -2.51 & iii or ms & - & + j150155.78 - 631332.5 & @xmath802 & @xmath803 & @xmath804 & @xmath805 & @xmath806 & @xmath807 & @xmath650 & -2.71 & iii or ms & - & + j150156.33 - 632315.1 & @xmath808 & @xmath809 & @xmath98 & @xmath810 & @xmath811 & @xmath812 & @xmath813 & -1.93 & iii or ms & - & + j150156.59 - 631728.2 & @xmath814 & @xmath815 & @xmath816 & @xmath817 & @xmath818 & @xmath819 & @xmath238 & -1.39 & ii & -0.13 & + j150159.36 - 631159.8 & @xmath525 & @xmath820 & @xmath821 & @xmath822 & @xmath823 & @xmath824 & @xmath825 & -0.4 & ii & -0.51 & + j150203.14 - 630458.3 & @xmath826 & @xmath827 & @xmath828 & @xmath829 & @xmath830 & @xmath831 & @xmath832 & -2.25 & iii or ms & - & + j150203.82 - 625442.2 & @xmath833 & @xmath834 & @xmath835 & @xmath836 & @xmath837 & @xmath838 & @xmath727 & -2.69 & iii or ms & - & + j150205.29 - 630714.5 & @xmath839 & @xmath840 & @xmath841 & @xmath842 & @xmath843 & @xmath844 & @xmath845 & -2.54 & iii or ms & - & + j150206.96 - 632047.7 & - & - & - & @xmath846 & @xmath847 & @xmath644 & @xmath848 & 2.45 & i & - & + j150208.37 - 630332.6 & @xmath849 & @xmath850 & @xmath851 & @xmath852 & @xmath853 & @xmath317 & @xmath854 & -0.89 & ii & -0.48 & + j150212.23 - 631525.6 & - & @xmath855 & @xmath493 & @xmath856 & @xmath857 & @xmath858 & @xmath859 & 1.03 & i & - & iras
14580 - 6303 ; flow i +
j150215.35 - 632028.3 & @xmath529 & @xmath860 & @xmath160 & @xmath861 & @xmath862 & @xmath86 & @xmath863 & 0.62 & i & 0.01 & + j150215.97 - 631749.2 & @xmath864 & @xmath865 & @xmath866 & @xmath867 & @xmath868 & @xmath869 & @xmath870 & -2.34 & iii or ms & - & + j150216.19 - 633738.2 & @xmath871 & @xmath872 & @xmath873 & @xmath874 & @xmath875 & @xmath876 & @xmath877 & -2.78 & iii or ms & - & + j150217.04 - 633413.5 & @xmath878 & @xmath879 & @xmath880 & @xmath881 & @xmath882 & @xmath883 & @xmath884 & -2.71 & iii or ms & - & + j150221.83 - 631301.9 & @xmath885 & @xmath886 & @xmath887 & @xmath888 & @xmath889 & @xmath230 & @xmath890 & -2.28 & iii or ms & - & + j150222.43 - 630458.4 & @xmath891 & @xmath892 & @xmath893 & @xmath894 & @xmath895 & @xmath896 & @xmath897 & -2.34 & iii or ms & - & + j150226.71 - 630511.6 & @xmath898 & @xmath899 & @xmath900 & @xmath901 & @xmath902 & @xmath903 & @xmath904 & -1.85 & iii or ms & - & + j150227.87 - 631843.3 & @xmath401 & @xmath905 & @xmath906 & @xmath907 & @xmath908 & @xmath909 & @xmath910 & -0.7 & ii & -0.32 & + j150227.91 - 630639.2 & @xmath455 & @xmath488 & @xmath911 & @xmath912 & @xmath175 & @xmath580 & @xmath80 & -0.2 & flat & -0.93 & + j150229.62 - 631738.1 & @xmath913 & @xmath914 & @xmath915 & @xmath916 & @xmath917 & @xmath918 & @xmath919 & -2.1 & iii or ms & - & + j150236.02 - 632623.8 & @xmath920 & @xmath921 & @xmath922 & @xmath923 & @xmath924 & @xmath925 & @xmath926 & -2.55 & iii or ms & - & + j150236.36 - 632131.1 & - & - & - & @xmath927 & @xmath928 & @xmath929 & @xmath930 & 1.14 & i & - & + j150236.87 - 633836.0 & @xmath931 & @xmath932 & @xmath933 & @xmath934 & @xmath935 & @xmath936 & @xmath937 & -2.83 & iii or ms & - & + j150238.02 - 631900.3 & - & - & - & @xmath938 & @xmath529 & 0.32u & @xmath939 & 1.66 & i & - & very red ; [ 4.6]-[22]=6.60 + j150238.90 - 631122.4 & @xmath940 & @xmath941 & @xmath942 & @xmath943 & @xmath511 & @xmath944 & @xmath945 & -0.35 & ii & 0.39 & + j150243.24 - 631112.2 & @xmath946 & @xmath947 & @xmath894 & @xmath948 & @xmath949 & @xmath950 & @xmath951 & -2.11 & iii or ms & - & + j150244.12 - 631210.0 & @xmath927 & @xmath952 & @xmath953 & @xmath954 & @xmath518 & @xmath955 & @xmath956 & -0.24 & flat & -0.98 & + j150245.32 - 630905.5 & @xmath957 & @xmath958 & @xmath959 & @xmath960 & @xmath961 & @xmath962 & @xmath963 & -2.54 & iii or ms & - & + j150245.86 - 625850.4 & @xmath964 & @xmath965 & @xmath966 & @xmath967 & @xmath968 & @xmath969 & @xmath970 & -2.84 & iii or ms & - & + j150250.84 - 631252.8 & @xmath971 & @xmath435 & @xmath911 & @xmath972 & @xmath973 & @xmath519 & @xmath974 & -0.42 & ii & -1.08 & + j150254.68 - 631301.8 & @xmath975 & @xmath976 & @xmath977 & @xmath978 & @xmath979 & @xmath980 & @xmath981 & -1.84 & iii or ms & - & + j150259.25 - 633713.1 & @xmath982 & @xmath983 & @xmath984 & @xmath985 & @xmath986 & @xmath987 & @xmath988 & -2.77 & iii or ms & - & + j150303.32 - 625443.7 & @xmath989 & @xmath990 & @xmath735 & @xmath991 & @xmath992 & @xmath993 & @xmath994 & -2.33 & iii or ms & - & + j150305.02 - 630912.2 & @xmath995 & @xmath996 & @xmath997 & @xmath998 & @xmath999 & @xmath1000 & @xmath1001 & -2.71 & iii or ms & - & + j150309.26 - 625821.1 & 1.44u & @xmath1002 & 2.14u & @xmath490 & @xmath1003 & @xmath1004 & @xmath1005 & 0.14 & flat & - & + j150309.36 - 632428.0 & @xmath1006 & @xmath1007 & @xmath569 & @xmath1008 & @xmath693 & @xmath1009 & @xmath1010 & -0.65 & ii & -0.35 & mo94 - 12 + j150309.47 - 631640.1 & @xmath1011 & @xmath1012 & @xmath46 & @xmath1013 & @xmath1014 & @xmath231 & @xmath1015 & -2.11 & iii or ms & - & + j150310.36 - 632358.2 & @xmath1016 & @xmath1017 & @xmath1018 & @xmath1019 & @xmath1020 & @xmath1021 & @xmath1022 & -2.83 & iii or ms & - & + j150311.55 - 631051.9 & @xmath1023 & @xmath1024 & @xmath1025 & @xmath1026 & @xmath1027 & @xmath1028 & @xmath1029 & -1.92 & iii or ms & - & iras 14590 - 6259 + j150313.83 - 632146.0 & @xmath860 & @xmath541 & @xmath530 & @xmath911 & @xmath1030 & @xmath769 & @xmath1031 & -0.57 & ii & -0.84 & + j150314.93 - 631806.6 & @xmath1032 & @xmath1033 & @xmath1034 & @xmath1035 & @xmath1036 & @xmath1037 & @xmath1038 & -2.15 & iii or ms - & + j150323.80 - 632258.8 & @xmath1039 & @xmath1040 & @xmath1041 & @xmath450748 & @xmath4501129 & @xmath4502138 & @xmath4504651 & @xmath4500.0 & flat ?
& - & c ; vdbh65b + j150325.16 - 631122.1 & @xmath1042 & @xmath1043 & @xmath1044 & @xmath1045 & @xmath1046 & @xmath1047 & @xmath1048 & -0.5 & ii & 0.58 & + j150326.64 - 632410.9 & @xmath1049 & @xmath1050 & @xmath1051 & @xmath1052 & @xmath291 & @xmath1053 & @xmath1054 & -2.17 & iii or ms & - & + j150328.03 - 632335.4 & @xmath1055 & @xmath1056 & @xmath1057 & @xmath45011 & @xmath45016 & @xmath45023 & @xmath450149 & @xmath4500.1 & flat ? & - & c ; vdbh65b region + j150328.78 - 632316.4 & 0.20u & @xmath528 & @xmath1058 & @xmath4503 & @xmath45024 & @xmath450113 & @xmath450686 & @xmath4501.6 & i ? & - & c ; vdbh65b region + j150328.90 - 632209.0 & @xmath1059 & @xmath1060 & @xmath1061 & @xmath1062 & @xmath1063 & @xmath1064 & @xmath1065 & -2.22 & iii or ms & - & + j150329.09 - 630858.6 & @xmath1066 & @xmath1067 & @xmath1068 & @xmath1069 & @xmath1070 & @xmath1071 & @xmath1072 & -2.81 & iii or ms & - & + j150332.29 - 632356.3 & @xmath927 & @xmath1073 & @xmath1074 & @xmath1075 & @xmath1076 & @xmath1077 & @xmath1078 & -0.15 & flat & 0.0 & + j150332.94 - 630948.1 & @xmath1079 & @xmath1080 & @xmath1081 & @xmath1082 & @xmath179 & @xmath1083 & @xmath1084 & -2.07 & iii or ms & - & + j150335.29 - 632250.4 & @xmath524 & @xmath1085 & @xmath1086 & @xmath1087 & @xmath1088 & @xmath1076 & @xmath1089 & 0.27 & flat & -0.40 & + j150341.56 - 625918.8 & @xmath1090 & @xmath1091 & @xmath1092 & @xmath1093 & @xmath1094 & @xmath1095 & @xmath1096 & -2.68 & iii or ms & - & + j150342.13 - 631936.7 & @xmath1097 & @xmath1098 & @xmath1099 & @xmath1100 & @xmath1101 & @xmath1102 & @xmath45 & -2.65 & iii or ms & - & + j150343.90 - 630035.7 & @xmath1103 & @xmath1104 & @xmath1105 & @xmath1106 & @xmath1107 & @xmath726 & @xmath1108 & -2.14 & iii or ms & - & + j150345.69 - 632341.2 & @xmath1109 & @xmath470 & @xmath1110 & @xmath1111 & @xmath1112 & @xmath1113 & @xmath1114 & -0.46 & ii & -0.31 & + j150348.08 - 633207.2 & @xmath1115 & @xmath1116 & @xmath1117 & @xmath1118 & @xmath1119 & @xmath1120 & @xmath1121 & -0.33 & ii & 0.02 & iras 14596 - 6320 ; flow h + j150348.39 - 632632.3 & - & - & - & 0.04u & @xmath1122 & 0.23u & @xmath1123 & - & n / a & - & very red ; [ 4.6]-[22]=8.48 + j150348.40 - 631435.8 & @xmath1124 & @xmath1125 & @xmath230 & @xmath675 & @xmath1126 & @xmath1127 & @xmath1128 & -0.79 & ii & -0.26 & + j150348.97 - 632343.5 & @xmath524 & @xmath143 & @xmath1129 & @xmath162 & @xmath1130 & @xmath1131 & @xmath1132 & 0.23 & flat & -0.12 & + j150353.05 - 632554.8 & - & - & - & @xmath528 & @xmath555 & @xmath521 & @xmath1133 & 1.29 & i & - & + j150355.14 - 633122.6 & @xmath1134 & @xmath1135 & @xmath1136 & @xmath1137 & @xmath1138 & @xmath1139 & @xmath1140 & -2.56 & iii or ms & - & + j150404.59 - 625739.1 & @xmath1141 & @xmath1142 & @xmath1143 & @xmath1144 & @xmath1145 & @xmath1146 & @xmath1147 & -2.83 & iii or ms & - & + j150405.88 - 632031.8 & @xmath1148 & @xmath1149 & @xmath1150 & @xmath1151 & @xmath1152 & @xmath1153 & @xmath1154 & -2.65 & iii or ms & - & + j150407.64 - 625835.0 & @xmath1155 & @xmath1156 & @xmath1157 & @xmath1158 & @xmath1159 & @xmath1160 & @xmath1161 & -2.83 & iii or ms & - & + j150410.07 - 630535.0 & @xmath1162 & @xmath1163 & @xmath1164 & @xmath1165 & @xmath1166 & @xmath1167 & @xmath1168 & -2.75 & iii or ms & - & + j150410.25 - 625405.5 & @xmath1169 & @xmath1170 & @xmath1171 & @xmath1172 & @xmath1173 & @xmath1174 & @xmath1175 & -2.35 & iii or ms & - & + j150414.17 - 630108.0 & @xmath1176 & @xmath1177 & @xmath1178 & @xmath1179 & @xmath1180 & @xmath1181 & @xmath1182 & -2.1 & iii or ms & - & + j150414.73 - 633634.0 & @xmath235 & @xmath620 & @xmath1183 & @xmath1184 & @xmath1185 & @xmath1186 & @xmath1187 & -0.54 & ii & -0.46 & + j150414.83 - 632425.4 & @xmath1188 & @xmath1189 & @xmath1190 & @xmath1191 & @xmath1192 & @xmath987 & @xmath1193 & -2.67 & iii or ms & - & + j150415.72 - 633406.7 & @xmath1100 & @xmath1194 & @xmath1195 & @xmath1196 & @xmath1197 & @xmath1198 & @xmath1199 & -2.7 & iii or ms & - & + j150416.50 - 631210.4 & @xmath1200 & @xmath1201 & @xmath1202 & @xmath1203 & @xmath1204 & @xmath1205 & @xmath1206 & -2.85 & iii or ms & - & + [ tab - sources ] | the _ wide - field infrared survey explorer _ has uncovered a population of young stellar objects in the western circinus molecular cloud .
images show the ysos to be clustered into two main groups that are coincident with dark filamentary structure in the nebulosity .
analysis of photometry shows numerous class i and ii objects .
the locations of several of these objects are found to correspond to known dense cores and co outflows .
class i objects tend to be concentrated in dense aggregates , and class ii objects more evenly distributed throughout the region . |
when a sediment bed is exposed to a fluid flow , particles can be entrained and transported by different mechanisms .
the transport regime depends primarily on the inertial characteristics of the particles and the fluid .
sufficiently light particles are transported as suspended load in which their weight is supported by the turbulence of the fluid .
in contrast , particles which are sufficiently heavy are transported along the surface @xcite .
this type of transport incorporates two main transport modes , namely : _
saltation _ ,
which consists of sediment grains jumping downstream close to the ground at nearly ballistic trajectories , and _ creep _ , which consists of particles rolling and sliding along the sediment bed .
sediment transport along the surface is responsible for a wide range of geophysical phenomena , including surface erosion , dust aerosol emission , and the formation and migration of dunes @xcite .
therefore , the quantitative understanding of sediment transport may improve our understanding of river beds evolution @xcite , the emission of atmospheric dust @xcite and the dynamics of planetary sand landscapes @xcite .
once sediment transport begins , the fluid loses momentum to accelerate the particles as a consequence of newton s second law ( the transport - flow feedback , e.g. @xcite ) .
therefore , the sediment flux , @xmath0 , which is the average momentum of grains transported per unit soil area , is limited by an equilibrium value , the saturated flux , @xmath1 .
although previous studies focused on this equilibrium flux ( e.g. @xcite ) , the dynamics of sediment landscapes is controlled by situations _ out - of - equilibrium_. in particular , the sediment flux needs a spatial lag the so - called saturation length , @xmath2 to adapt to a change in flow conditions @xcite .
this saturation length introduces the main relevant length - scale in the dynamics of sediment landscapes under water and on the surface of planetary bodies .
for instance , the saturation length controls the minimal size of crescent - shaped ( barchan ) dunes moving on top of bedrock , as well as the wavelength of the smallest dunes ( the so - called `` elementary dunes '' ) emerging on top of a sediment bed @xcite .
although important insights were gained recently from experimental studies @xcite , the physics behind the saturation length , and thus the dependence of @xmath2 on flow and sediment attributes , is still insufficiently understood .
one of the most important deficiencies in our understanding of the dependence of @xmath2 on flow and sediment conditions is that it remains uncertain which mechanisms are most important in determining the saturation of the sediment mass flux . on the one hand
, it has been suggested that the acceleration of transported particles due to fluid drag is the dominant relaxation mechanism @xcite .
this model neglects the entrainment of sediment bed particles due to fluid lift , as well as the entrainment of sediment bed particles and the deceleration of transported particles due to collisions of transported particles with the sediment bed ( grain - bed collisions ) . on the other hand , the entrainment of sediment bed particles by fluid
lift and grain - bed collisions has also been proposed to be the dominant relaxation mechanisms @xcite
. however , these models neglect momentum changes of transported particles , which is exactly the opposite situation of the models in refs . @xcite .
moreover , to our knowledge , all previous models neglected a further relaxation mechanism of the sediment flux , namely , the relaxation of the fluid speed in the transport layer ( @xmath3 ) due to the saturation of the transport - flow feedback @xcite . to address this situation and develop an accurate expression for @xmath2 that can be used in future studies
, this paper presents a model for flux saturation in sediment transport which , for the first time , accounts for _ all _ aforementioned mechanisms for the saturation of sediment flux .
in particular , our theoretical model accounts for the coupling between the entrainment of sediment bed particles due to fluid lift and grain - bed collisions , the acceleration and deceleration of transported particles due to fluid forces and grain - bed collisions , and the saturation of @xmath3 due to the saturation of the transport - flow feedback . our analytical model allows us to derive a closed expression for @xmath2 which can be applied to different physical environments .
our model suggests that grain - bed collisions , which have been neglected in all previous studies , have an important influence on the saturation length , @xmath2 .
moreover , our model suggests that the relaxation of @xmath3 plays an important role for sediment transport in dilute fluids ( aeolian transport ) , whereas it plays a negligible role for sediment transport in dense fluids ( subaqueous transport ) . in a recent letter ( see ref .
@xcite ) , we presented our equation for @xmath2 and showed that it is consistent with measurements of @xmath2 in both subaqueous and aeolian sediment transport regimes over at least five orders of magnitude in the ratio between fluid and particle density .
in the present paper , we derive the analytical model presented in ref .
@xcite in more detail and study the properties of the equations governing the behavior of the saturation length in both transport regimes . since ref .
@xcite includes a detailed comparison of our model against measurements , no model comparisons against measurements are included here .
this paper is organized as follows .
sections [ derivation ] and [ obtaining_ls ] discuss the analytical treatment of flux saturation . in the former section , we derive the mass and momentum conservation equations for the layer of sediments in transport , as well as the differential equation of the sediment flux in terms of the mass density and average velocity of the transported particles .
these equations allow us to obtain a mathematical expression for the saturation length of sediment transport , which is presented in section [ obtaining_ls ] .
this section also discusses how to determine the quantities appearing in the saturation length equation , which encode the attributes of sediment and flow , as well as the characteristics of sediment entrainment and particle - fluid interactions . in section [ discussion ]
we use our theoretical expression to perform a study of the saturation length as a function of the relevant physical quantities controlling saturation of sediment flux .
conclusions are presented in section [ conclusions ] .
the downstream evolution of the sediment flux , @xmath0 , towards its equilibrium value , @xmath1 , can be described by the following equation @xcite , which is identical to eq .
( 1 ) of ref .
@xcite , @xmath4 which is valid in the regime where @xmath0 is close to saturation ( @xmath5 ) .
the length - scale @xmath2 , the saturation length , characterizes the response of the sediment flux due to a small change in flow conditions around equilibrium . since @xmath6
, @xmath2 can be written as the negative inverse first - order taylor coefficient of @xmath7 , @xmath8 in this section , we derive the equations that describe the downstream evolution of the sediment mass flux , @xmath0 , towards its equilibrium value , @xmath9 , in sediment transport under turbulent boundary layer flow .
the mass flux @xmath0 is defined as @xmath10 , where @xmath11 is the average transported mass per unit soil area and @xmath12 is the average particle velocity .
therefore , the saturation of @xmath0 is dictated by the mechanisms governing the relaxation of @xmath11 and @xmath12 towards their saturated values , @xmath13 and @xmath14 , respectively .
the quantitative description of the saturation processes of @xmath11 and @xmath12 requires incorporation of all relevant forces acting on the sediment particles in transport , namely drag , gravity , buoyancy , collision forces between particles in transport ( `` mid - fluid collisions '' ) and friction due to collisions between particles and the bed . indeed , moraga et al .
@xcite found experimentally that lift forces due to shear flow acting on a particle surrounded by fluid which have often been assumed to be significant during transport ( e.g. @xcite ) are approximately an order of magnitude smaller than the drag force and can be , thus , neglected in our calculations . on the other hand ,
the so - called added mass force exerted by accelerated or decelerated particles to dislodge the fluid as they move through it leads to enhanced inertia of the particles in transport .
this added mass effect plays a relevant role for the motion of the particles @xcite , and thus we also take it into account .
our analytical treatment applies to situations where the fluid velocity is not too high such that only transport through saltation or creep ( the main transport modes of particles along the surface @xcite ) is considered .
transport through suspension or dense transport regimes , such as sheet flow @xcite , are , thus , not considered . in section [ definitions ]
we first present the definitions and notations used in our study . afterwards in section [ mass_and_momentum ] , we present the local conservation equations , from which we obtain the saturation equations , presented in section [ sec : saturation ] .
we use a three - dimensional coordinate system @xmath15 , where @xmath16 denotes the direction of fluid motion , @xmath17 is the lateral direction and @xmath18 is the vertical direction .
the top of the sediment bed , which corresponds to the height at which the local particle concentration equals approximately @xmath19 of the particle concentration deep within the bed @xcite , is located at the vertical position @xmath20 .
here we use the approximation that the slopes of bedforms are usually very small ( @xmath21 ) .
moreover , since the time - scale of the relaxation of the sediment flux due to changes in the flow is typically much smaller than the time - scale of the evolution of bedforms ( dunes and ripples ) @xcite ( @xmath22 ) , we can adopt the approximation that the transport over the sediment landscape is in the steady - state , i.e. @xmath23 , where @xmath24 denotes time .
furthermore , since our description relates to the saturation of the mass flux @xmath0 due to changes in the downstream direction , we consider a laterally invariant sediment bed ( @xmath25 ) .
we consider a certain microscopic configuration of @xmath26 particles ( including the limit @xmath27 ) labeled by an upper index @xmath28 whose centers of mass are located at @xmath29 .
each particle has a mass @xmath30 , a velocity @xmath31 , and is subjected to a force @xmath32 resulting in an acceleration @xmath33 .
these forces include both external body forces ( @xmath34 ) and interparticle contact forces . in general these forces
are non - conservative .
the interparticle contact forces occur for all pairs of contacting particles . we therefore denote them by @xmath35 , which is the contact force applied by the particle with the number @xmath28 on the particle with the number @xmath36 .
we note that @xmath37 if these particles are not in contact , and we define @xmath38 ( no self - interaction ) .
hence , the total acceleration of particle @xmath28 can be written as , @xmath39 we define @xmath40 , the density of a certain microscopic configuration of particles at time @xmath24 , as @xmath41 it describes the number of particles , @xmath42 , with positions , velocities , and masses in infinitesimal intervals around @xmath43 , @xmath44 , and @xmath36 , respectively , at time @xmath24 , @xmath45 moreover , @xmath46 determines the mass density , @xmath47 while the mass - weighted average of a quantity @xmath48 is defined through the equation , @xmath49 in eqs .
( [ def_rho ] ) and ( [ def_av ] ) @xmath50 denotes the time average , @xmath51 using these definitions , we can calculate the total transported mass per unit soil area ( @xmath11 ) , the total mass flux ( @xmath0 ) , and the average particle velocity ( @xmath12 ) from the expressions , @xmath52 respectively , where the overbar denotes the mass - weighted height average , @xmath53 in this section , the local average mass and momentum conservation equations for our particle system are presented using the notations and definitions introduced in the last section .
the derivation of these conservation equations can be found in babic @xcite .
for our system ( @xmath54 ) , these equations are , @xmath55 where @xmath56 is given by @xcite , @xmath57 with @xmath58 .
@xmath56 is the contact force contribution to the particle stress tensor since its gradient compensates the contact force density @xcite , @xmath59 it describes the momentum flux due to collisions between particles .
in fact , even though the total momentum is conserved in collisions , the finite size of the particles and thus @xmath60 lead to a shift of the location of this momentum .
we note that this shift of the momentum location in collisions has been neglected in our model derivation in ref .
@xcite ( dilute approximation ) . as a consequence , eq .
( [ mombx ] ) is a generalization of eq .
( 2 ) of ref . @xcite , such that these two equations are equal if the contributions from @xmath56 in eq .
( [ mombx ] ) are neglected .
the distribution @xmath61 appearing in eq .
( [ pij ] ) is the mathematical expression for a delta line between @xmath62 and @xmath29 .
integrating this distribution over an arbitrary domain yields the fraction of the line contained in this domain .
the inhomogeneities introduced by this and the other delta distributions indirectly appearing in quantities of the type @xmath63 are smoothed out by the time averaging procedure @xmath50 , which is also incorporated in the definition of @xmath64 .
the results of the last section can be now used in order to derive the saturation equations for the average transported mass per unit soil area ( @xmath11 ) and the average particle velocity ( @xmath12 ) , used to define the sediment flux , @xmath65 . to do so
, we first integrate eqs .
( [ massb])-([mombz ] ) over the height .
this calculation is the subject of section [ height_integration ] .
thereafter , in section [ momentum_balance ] , we combine the resulting horizontal and vertical momentum balances by means of a coulomb friction law and rewrite each term of the horizontal momentum balance equation in terms of @xmath11 and @xmath12 .
we then present the mass and horizontal momentum balance equations in their final form in section [ final_form ] .
since our description relates to the saturation of the mass flux @xmath0 due to changes in the downstream direction ( @xmath16 ) , we integrate eqs .
( [ massb])-([mombz ] ) over height ( @xmath66 ) . by using eqs .
( [ m])-([heightav ] ) and by further taking into account @xmath21 and @xmath67 , this height - integration yields , @xmath68 we note that eq .
( [ mombalance3x_new ] ) corresponds to eq .
( 2 ) of ref .
@xcite if the contributions from @xmath56 in eq .
( [ mombalance3x_new ] ) are neglected ( dilute approximation ) . _ coulomb friction law _ the terms @xmath69 and @xmath70 are the vertical fluxes of horizontal and vertical momentum component per unit volume at the location of the sediment bed , respectively , whereby the velocity terms are the contributions due to particle motion , and @xmath71 and @xmath72 are the contributions due to collisional momentum transfer .
in other words , these two terms describe the total amounts of horizontal and vertical momentum , respectively , per unit soil area that enter the transport layer per unit time from the sediment bed .
these momentum changes per unit area and time of the transport layer can be seen as being caused by an effective force per unit area ( @xmath73 ) which the sediment bed applies on the transport layer ,
@xmath74 bagnold @xcite was the first who proposed that these force components are related to each other through a coulomb friction law , independent of whether the transport regime is subaqueous or aeolian .
that is , @xmath75 where @xmath76 is the coulomb friction coefficient .
models for saturated sediment transport using this coulomb friction law have been successfully validated through comparison with experiments , thus giving support to the coulomb friction law adapted to sediment transport ( e.g. @xcite ) .
additional support comes from numerical simulations of saturated ( @xmath77 ) granular couette flows under gravity .
zhang and campbell @xcite found for such flows that the interface between the particle bed and the transport layer is characterized by a constant ratio between the @xmath78 and @xmath79 components of the particle stress tensor ( @xmath80 ) , @xmath81 .
since both couette flow and sediment transport along the surface are granular shear flows , it seems reasonable that also the interface between the sediment bed and the transport layer for saturated sediment transport along the surface is characterized by such a law .
indeed , @xmath82 and @xmath83 become equal to @xmath84 and @xmath85 , respectively , if @xmath86 @xcite , which is fulfilled for saturated sediment transport since @xmath77 implies @xmath87 ( cf .
( [ massb ] ) ) , which in turn implies @xmath86 due to @xmath88 vanishing sufficiently deep within the sediment bed .
finally , it seems reasonable that the coulomb friction law should be also approximately valid in situations weakly out - of - equilibrium @xcite , provided the sediment flux is close to its saturated value . assuming the validity of eq .
( [ coulombassumption ] ) , we can combine eqs .
( [ mombalance3x_new ] ) and ( [ mombalance3z ] ) to , @xmath89 where @xmath90 is a correlation factor given by , @xmath91 _ the correlation factor _
since we are only interested in situations close to equilibrium , and since at equilibrium @xmath86 ( see discussion in the previous paragraph ) , it follows @xmath92 ( @xmath76 is of order unity ) .
moreover , for sufficiently dilute granular flows , the momentum transfer in collisions is small and thus @xmath93 . while sediment transport in the aeolian regime is certainly dilute enough to ensure this condition for most of the transport layer , sediment transport in the subaqueous regime
might not fulfill it because a large part of the transport occurs in rather dense regions of the transport layer @xcite .
however , using the code of durn et al .
@xcite , we confirmed that @xmath93 also for subaqueous transport .
hence , @xmath90 can be approximated as , @xmath94 we confirmed , using the code of durn et al .
@xcite , that for transport in equilibrium ( @xmath77 ) @xmath90 is nearly constant with the fluid shear velocity , @xmath95 , in both sediment transport regimes .
hence , it seems reasonable that changes of @xmath90 with @xmath16 during the saturation process of the sediment flux close to equilibrium can be regarded as negligible compared to the corresponding changes of @xmath11 or @xmath12 with @xmath16 . in this manner
, we can consider the value of @xmath90 associated with sediment transport in equilibrium , independent of the downstream position and of the fluid shear velocity .
this leads to the following approximation for @xmath90 , @xmath96 where @xmath97 is the equilibrium value of @xmath98 .
this equilibrium value of @xmath90 can be determined from experiments as we will discuss in section [ cv_aeolian ] and [ cv_subaqueous ] .
now we express both terms on the right - hand - side of the momentum conservation equation , i.e. eq . ( [ mombalance3x ] ) , as functions of @xmath11 and @xmath12 in order to obtain a differential equation describing the saturation of @xmath11 and @xmath12 . the first term on the right - hand - side of eq .
( [ mombalance3x ] ) can be written as @xcite , @xmath99 where @xmath100 is the ratio between sediment and fluid density ; @xmath101 is defined as , @xmath102 which is the difference between the average fluid velocity ( @xmath103 ) and the average horizontal particle velocity ( @xmath12 ) , where @xmath104 is the fluid velocity profile ; @xmath105 is the drag coefficient , which is a function of @xmath101 , and @xmath106 accounts for the added mass force through , @xmath107 the added mass force arises when the particle is accelerated relative to the surrounding fluid , because the fluid layer immediately surrounding the particle will also be accelerated . as denoted by eq .
( [ ca ] ) , this `` added mass '' of the fluid layer amounts to approximately one half the weight of the fluid displaced by the particle @xcite . while the added mass correction is significant for transport in a dense medium such as water @xcite , it is negligibly small for sediment transport in the aeolian regime since @xmath108 for large @xmath109 .
thus , this correction is usually disregarded in studies of aeolian sediment transport ( e.g. refs .
we note that eq .
( [ drag ] ) is not valid for dense transport regimes like sheet flow , in which the drag coefficient displays a strong dependence on the concentration profile of transported particles @xcite . in this manner ,
( [ drag ] ) can be used in the present study because our analytical treatment considers the two main modes of transport , namely saltation and creep . the second term on the right - hand - side of eq .
( [ mombalance3x ] ) , @xmath110 , can be taken as approximately equal to the buoyancy - reduced gravity force @xcite corrected by the added mass force .
it can be written as , @xmath111 where @xmath112 is the buoyancy - reduced value of the gravity constant , @xmath113 . by substituting eqs .
( [ drag ] ) and ( [ friction ] ) into eq .
( [ mombalance3x ] ) using @xmath114 ( cf .
( [ cvdef2 ] ) ) , we obtain the momentum conservation equation in terms of @xmath11 and @xmath12 , whereas eq .
( [ massb2 ] ) gives the mass balance .
therefore , the mass and momentum conservation equations in their final form read , @xmath115 we note that eq .
( [ momconserv ] ) is identical to eq .
( 4 ) of ref .
@xcite if the definition of @xmath106 ( eq .
( [ ca ] ) ) is inserted .
we further note that eq .
( [ momconserv ] ) can be used to obtain the saturated value @xmath116 of the velocity difference @xmath101 . by using @xmath117 ( saturated sediment transport ) ,
we obtain , @xmath118 which can be numerically solved for @xmath116 .
in this section , we use the results presented in last section in order to derive a closed expression for the saturation length as a function of the attributes of sediment and flow , both for aeolian and subaqueous regimes .
the derivation of the saturation length equation is the subject of section [ derivation_of_ls ] . in section [ equations_for_ls ]
we present and discuss the resulting equation for the saturation length . in sections [ ls_aeolian ] and [ ls_subaqueous ]
we show how the saturation length equation can be applied to compute @xmath2 in the aeolian and subaqueous regimes , respectively .
close to equilibrium , @xmath11 and @xmath12 saturate simultaneously following a certain function @xmath119 , where @xmath120 .
this function is linked to the characteristics of the erosion and deposition of bed material and thus to the unknown shape of @xmath121 as a function of @xmath11 and @xmath12 in eq .
( [ massconserv ] ) . moreover , also the mean fluid velocity @xmath3 will saturate following a certain function @xmath122 close to the saturated regime , since @xmath3 is influenced by the feedback of the sediment transport on the fluid flow .
therefore , eq . ( [ vr ] ) becomes , @xmath123 by taking into account that both @xmath11 and @xmath3 are functions of @xmath12 , and by using eq .
( [ vrs ] ) , we can rewrite the momentum balance eq .
( [ momconserv ] ) in such a way to obtain the following expression for @xmath124 , @xmath125 , \label{omegadef}\ ] ] where the functions @xmath126 and @xmath127 are given by the equations , @xmath128 furthermore , since @xmath129 , we obtain , @xmath130 in this manner , using eq .
( [ omegadef ] ) , @xmath131 can be written as , @xmath132 using eqs .
( [ dv_dq ] ) and ( [ gamma_v ] ) , we can write eq .
( [ lsdef2 ] ) for the saturation length as , @xmath133 where we further used that @xmath134 . calculating @xmath2 through eq .
( [ lv ] ) requires obtaining an expression for @xmath135 , where @xmath136 is defined in eq .
( [ omegadef ] ) .
however , @xmath136 incorporates , through the function @xmath127 defined in eq .
( [ b_v ] ) , a dependence on the equilibrium value of the relative velocity @xmath101 , i.e. @xmath116 . in order to obtain an expression for @xmath116 , we solve eq .
( [ vrs ] ) for @xmath116 using the drag law of julien @xcite for natural sediment , which writes , @xmath137 whereas we find that the specific choice of the drag law has only a small effect on the value of @xmath2 obtained from our calculations . by substituting the expression for @xmath138 , obtained with eq .
( [ dragjulien ] ) , into eq .
( [ vrs ] ) , and solving this equation for @xmath116 , yields , @xmath139 this equation is , then , used to compute @xmath140 through eq .
( [ b_v ] ) , whereupon @xmath136 can be obtained using eqs .
( [ vrv ] ) and ( [ omegadef ] ) . the resulting expression for the saturation length , computed with eq .
( [ lv ] ) , reads , @xmath141 where the quantity , @xmath142 describes the relative change of @xmath11 with @xmath12 close to the saturated regime , while @xmath143 is given by , @xmath144 and thus @xmath143 encodes information about the drag law . in order to obtain our final expression for @xmath2 ,
we need to express @xmath145 .
we note that , for the saturated state , the mean flow velocity @xmath3 is dominantly a function of the shear velocity @xmath95 and the shear velocity at the bed @xcite , that is , @xmath146 the shear velocity at the bed , @xmath147 , is reduced due to the feedback of the sediment transport on the fluid flow , where @xmath148 is the fluid shear stress profile
. we can express @xmath147 using the inner turbulent boundary layer approximation of the navier - stokes equations .
these equations approximate the navier - stokes equations for heights @xmath18 much smaller than the height @xmath149 of the boundary layer , which is the region in which we are interested .
george @xcite derived the inner boundary layer approximation of the navier - stokes equations in the absence of an external body force . in the presence of an external body force
, these equations must be slightly modified by adding the body force term in the momentum equations .
the horizontal momentum equation thus writes @xcite , @xmath150 where @xmath151 is the horizontal body force per unit volume acting on the flow at each height @xmath18 .
@xmath152 is the drag force per unit volume which the particles apply on the fluid . in other words , @xmath152 is the reaction force per unit volume of the horizontal force per unit volume which the fluid applies on the particles .
that is , @xmath153 we then substitute eq .
( [ fxbody ] ) into eq .
( [ boundarylayer ] ) and integrate this equation from @xmath154 to @xmath155 , where @xmath156 is a height which incorporates the entire transport layer , thereby using @xmath157 , and @xmath158 .
this leads to , @xmath159 by substituting this equation into eq .
( [ ub ] ) and using eq .
( [ drag ] ) , we obtain the following equation for @xmath147 , @xmath160 since @xmath95 does not depend on @xmath12 , we can now express @xmath145 as , @xmath161 where @xmath162 is the value of @xmath147 in equilibrium , and the quantity @xmath163 is given by the equation , @xmath164 where we used , @xmath165 .
we note that @xmath163 describes the relative change of @xmath3 with @xmath147 close to the saturated regime .
moreover , the derivative @xmath166 can be calculated using eq .
( [ ubv ] ) with , @xmath167 , @xmath168 , and , @xmath169 which follows from @xmath170 .
we thus obtain , @xmath171.\end{aligned}\ ] ] this expression is substituted into eq .
( [ uvs ] ) , whereas the resulting equation is then solved for @xmath172 this is the term involving @xmath173 which we need to compute @xmath2 in eq .
( [ lvfinal ] ) . in this manner
, we finally obtain a closed expression for the saturation length , which we present and discuss in the next subsection . the equation for the saturation length , @xmath2 , which is identical to eq .
( 5 ) of ref .
@xcite if the definition of @xmath106 ( eq .
( [ ca ] ) ) is inserted , reads , @xmath174 where @xmath116 and @xmath143 are calculated using eqs .
( [ vrs ] ) and ( [ cd_fx ] ) , respectively .
in addition , the last factor @xmath175 on the right - hand - side of eq .
( [ lvfinal2 ] ) is given by the equation , @xmath176\cdot\left(\dfrac{u_{\ast}^2}{u_{bs}^2}-1\right)}{1+\left[\dfrac{c_uc_m\cdot(v_s+v_{rs})}{2v_s}\right]\cdot\left(\dfrac{u_{\ast}^2}{u_{bs}^2}-1\right)}\approxeq\dfrac{1+\left[\dfrac{c_u\cdot(v_s+v_{rs})}{2fv_{rs}}\right]\cdot\left(\dfrac{u_{\ast}^2}{u_{{\mathrm{t}}}^2}-1\right)}{1+\dfrac{c_uc_m\cdot(v_s+v_{rs})}{2v_s}\cdot\left(\dfrac{u_{\ast}^2}{u_{{\mathrm{t}}}^2}-1\right ) } , \label{lsfluid}\ ] ] and @xmath175 thus encodes information about the saturation of the transport - flow feedback .
in fact , if the saturation of the transport - flow feedback is neglected ( @xmath177 ) , it follows @xmath178 and thus @xmath179 .
we note that eq .
( [ lsfluid ] ) is identical to eq .
( 9 ) of ref .
@xcite , which we obtained for aeolian transport , for @xmath180 ( see section [ ls_aeolian ] ) .
moreover , for transport in the subaqueous regime , @xmath181 as shown in section [ ls_subaqueous ] .
therefore , in this regime , eq . ( [ lsfluid ] ) gives @xmath182 , which is the result we obtained for subaqueous transport in ref .
in fact , using the corresponding values for @xmath183 and @xmath163 and inserting eq .
( [ ca ] ) , eq .
( [ lvfinal2 ] ) becomes equal to , @xmath184 for subaqueous transport and , @xmath185 for aeolian transport , where we further used @xmath186 for aeolian transport .
( [ lvfinal2subq ] ) and ( [ lvfinal2aeol ] ) are identical to eqs .
( 8) and ( 10 ) of ref .
@xcite , respectively . in eq .
( [ lsfluid ] ) , we assumed that the saturated shear velocity at the bed ( @xmath162 ) and the bed shear stress in equilibrium ( @xmath187 ) approximately equal @xmath188 and @xmath189 , respectively , i.e. the minimal shear velocity and the minimal shear stress at which sediment transport can be sustained , @xmath190 in the following , we present arguments which justify this assumption . for aeolian sediment transport , eqs .
( [ tauftaut ] ) and ( [ ubut ] ) are known as `` owen s hypothesis '' .
these equations are known to be approximately valid when @xmath95 is close to the threshold ( e.g. figure 2.10 in ref .
however , as @xmath95 increases , @xmath162 actually decreases away from @xmath188 @xcite .
nonetheless , the approximation which we use in eq .
( [ lsfluid ] ) is reasonable even for large shear velocities , since , when @xmath95 is significantly larger than @xmath188 ( which means @xmath191 for earth conditions with @xmath192 ) , we have that , @xmath193 which is nearly independent of @xmath162 . using this approximation with @xmath194 , eq .
( [ lvfinal2aeol ] ) becomes , @xmath195 which is identical to eq .
( 39 ) of the supplementary material of ref .
@xcite . for subaqueous sediment transport , eqs .
( [ tauftaut ] ) and ( [ ubut ] ) are known as `` bagnold s hypothesis '' .
this hypothesis is widely used in the literature ( e.g. @xcite ) , although some studies have questioned it ( e.g. @xcite ) .
however , there is evidence from recent studies that this hypothesis is approximately fulfilled . in order to review this evidence , we use eqs .
( [ drag ] ) , ( [ vrs ] ) , and ( [ tautauf ] ) to express @xmath13 as , @xmath196 . \label{ms}\end{aligned}\ ] ] to our knowledge , the only study in which @xmath13 has been measured as a function of @xmath197 is the recent study of lajeunesse et al .
@xcite , who obtained , using video - imaging techniques , that , @xmath198 .
\label{msshape}\end{aligned}\ ] ] therefore , if we assume @xmath199 as in eq .
( [ tauftaut ] ) , then , by comparing eqs .
( [ ms ] ) and ( [ msshape ] ) with @xmath200 valid for subaqueous sediment transport ( cf .
( [ ca ] ) with @xmath201 ) , we obtain @xmath202 and thus @xmath203 .
indeed , values within the range between @xmath204 and @xmath205 and thus consistent with the value of @xmath76 estimated above have been reported from measurements of particle trajectories in the subaqueous sediment transport @xcite .
further evidence that bagnold s hypothesis is approximately correct was provided by the recent numerical study of durn et al .
these authors simulated the dynamics of both the sediment bed and of transported particles at the single particle scale .
durn et al .
@xcite found that @xmath206 , which is similar to eq .
( [ msshape ] ) and can satisfactorily explain all simulated data with a single proportionality constant
. moreover , the authors also found that @xmath207 reduces to @xmath189 at a height @xmath18 very close to the top of the bed , @xmath208 .
given these separate lines of evidence , we believe that bagnold s hypothesis is a reasonable approximation .
moreover , we emphasize that our analysis for subaqueous sediment transport is not affected by this approximation , since we estimate in section [ cu_subaqueous ] that @xmath181 and thus @xmath182 regardless of the value of @xmath162 . in summary ,
the saturation length of sediment flux , @xmath2 , can be calculated using eq .
( [ lvfinal2 ] ) , where @xmath116 and @xmath143 are given by eqs .
( [ vrs2 ] ) and ( [ cd_fx ] ) , respectively , while eq .
( @xmath209 ) is used to compute the term , @xmath172 , which appears on the right - hand - side of eq .
( [ lvfinal2 ] ) .
these equations include certain quantities which depend on the characteristics of the sediment transport and thus on the transport regime .
these quantities are @xmath90 , @xmath183 , @xmath163 , @xmath76 , the saturated particle velocity @xmath14 , and the threshold shear velocity , @xmath188 .
we estimate these quantities for the aeolian regime of sediment transport in section [ ls_aeolian ] and for the subaqueous regime in section [ ls_subaqueous ] . in this section
, we estimate the parameters @xmath90 , @xmath163 , @xmath183 , @xmath76 , and express the saturated particle velocity @xmath14 and the threshold shear velocity @xmath188 for aeolian sediment transport . note that we estimate these parameters only roughly , which is sufficient in the light of the large scatter ( factor @xmath210 ) of the experimental data @xcite . in this section ,
we reiterate some of the results we obtained in section a1 of the supplementary material of ref .
the parameter @xmath90 ( eq .
( [ cvdef ] ) ) occurs as a prefactor in eq .
( [ lvfinal2 ] ) , and thus determines the magnitude of @xmath2 . since , @xmath211
we conclude that @xmath90 must be larger than unity , that is , @xmath212 however , experiments on aeolian sediment transport @xcite show that the change of @xmath213 with @xmath18 is small close to @xmath214 , where most of the transport takes place .
consequently , the value of @xmath90 must be close to unity .
we estimate @xmath90 from experiments on aeolian sediment transport @xcite .
creyssels et al .
@xcite measured an exponentially decaying particle concentration profile , @xmath215 and a linearly increasing particle velocity profile , @xmath216 where @xmath217 mm , @xmath218 m/s , and @xmath219 were not varying much with @xmath220 . using these measurements , we obtain , @xmath221 in order to obtain @xmath90 , it remains to estimate @xmath222 . in order to do so , we use measurements of greeley et al .
13 of ref .
@xcite ) , who reported a histogram of the horizontal particle velocity of the particles located at a height @xmath223 , from which we obtain , @xmath224 since the shape of the distribution of the horizontal particle velocity does not vary much with the height @xcite , we thus estimate @xmath90 as , @xmath225 in contrast to @xmath90 , the parameter @xmath163 ( eq . ( [ cudef ] ) ) significantly influences the functional shape of @xmath2 as a function of @xmath95 .
@xmath163 characterizes the significance of the transport - flow feedback for the saturation of the sediment flux .
for instance , @xmath178 means that the transport - flow feedback does not affect the saturation process since the flow is already saturated ( @xmath177 from eq .
( [ cudef ] ) ) . in order to estimate @xmath163 , we need to know how the mean fluid speed @xmath3 behaves as a function of the feedback - reduced bed shear velocity @xmath147 ( see eq .
( [ cudef ] ) ) . for aeolian sediment transport ,
the fluid speed is strongly suppressed by the reaction drag forces which the transported grains apply on the wind .
the feedback is , in fact , so strong that the mean fluid speed in the transport layer changes only weakly with @xmath95 @xcite . in leading - order approximation ,
the mean fluid speed is thus proportional to @xmath147 , @xmath226 we thus obtain , from eq .
( [ cudef ] ) , @xmath227 we note that a value of @xmath163 close to unity is obtained even if the more complicated dependence of @xmath3 on @xmath147 , obtained from modeling saturated sediment flux @xcite , is taken into account .
( [ eq : cu_value ] ) is approximately valid for @xmath228 @xcite . beyond this range
, turbulence - induced fluctuations of the shear velocity , neglected in the present work , should affect the value of @xmath163 .
the parameter @xmath183 , given by eq .
( [ cmdef ] ) ) , occurs as a prefactor in eq .
( [ lvfinal2 ] ) and it further affects the functional shape of @xmath229 since @xmath183 itself affects the feedback term , @xmath230 .
@xmath183 encodes the relative importance of the respective relaxation processes @xmath231 and @xmath232 for the saturation of the sediment flux .
there are two extreme cases : @xmath233 and @xmath234 .
the case @xmath235 means that the saturation of @xmath11 towards @xmath13 is much faster than the saturation of @xmath12 towards @xmath14 , while the opposite situation corresponds to the case @xmath234 . in order to estimate @xmath183 for the aeolian regime of sediment transport ,
we first estimate how the function @xmath119 behaves close to the saturated regime . for this purpose , we make use of the fact that , for aeolian sediment transport , the dominant mechanism which brings grains of the sediment bed into motion is the ejection of bed grains due to impacts of already transported grains , a mechanism known as `` splash '' ( see e.g. @xcite ) . it is known that ejection of new grains is mainly due to the impacts of the fastest transported particles , whereas the impacts of slow particles have a negligible effect on the splash process @xcite .
indeed , the speed of a fast impacting grain mainly determines the number of ejected grains , but not their ejection velocities , as found in experiments @xcite .
the ejected particles are typically slow compared to the rebound speed of the impacting particle . in other words , the impact of a fast grain naturally results in two species of particles : a single ( fast ) rebounding particle and many ejected ( slow ) particles . using numerical simulations of splash and particle trajectories
, andreotti @xcite could observe these two distinct species in the characteristics of transported particles .
the author noted that the slow species ( `` reptons '' ) accounts for the majority of transported mass per unit soil area ( @xmath11 ) .
furthermore , the author s analysis suggested that the impact flux of reptons , and thus , in good approximation , the total transported mass @xmath11 , adjusts to changes of the impact flux of the fast species ( `` saltons '' ) within a distance much shorter than @xmath2 .
therefore , it seems reasonable to treat @xmath11 as locally equilibrated with respect to the impact flux of saltons .
the locally equilibrated value @xmath236 of @xmath11 is proportional to the number of ejected particles per impact , which in turn is proportional to the impact speed of saltons @xcite and thus approximately proportional to @xmath12 .
a rough estimate of the function @xmath119 is therefore , @xmath237 which yields , @xmath238 the coulomb friction coefficient , @xmath76 , occurs as a prefactor in eq .
( [ lvfinal2 ] ) , and also changes the functional shape of @xmath2 through eq .
( [ vrs2 ] ) .
@xmath76 can be determined indirectly from measurements of the saturated mass of transported particles @xmath13 as a function of the shear velocity @xmath95 , which fulfills the equation @xcite , @xmath239 , \label{mudef}\end{aligned}\ ] ] note that this equation is eq .
( [ ms ] ) with @xmath199 ( eq .
( [ tauftaut ] ) ) . the value
, @xmath240 was found in a previous work @xcite through determining @xmath76 indirectly both from experiments as mentioned above and from numerical simulations of aeolian sediment transport in equilibrium .
the saturated particle velocity @xmath14 is dominantly controlling the dependence of @xmath2 on @xmath95 in eq .
( [ lvfinal2 ] ) .
since the dependence of @xmath14 on @xmath95 is rather weak for aeolian sediment transport @xcite , the saturation length @xmath2 will not change much with @xmath95 .
here we use an expression for @xmath14 which has been obtained in a recent work @xcite , since the values of saturated sediment flux obtained using this equation produced excellent quantitative agreement with measurements @xcite .
the expression for @xmath14 reads @xcite , @xmath241 where @xmath242 and @xmath243 are given by the equations , @xmath244 in these equations , @xmath245 is the exponential integral function , @xmath246 is the von krmn constant , and @xmath247 is a dimensional parameter encoding the influence of cohesion , while @xmath248 and @xmath249 are empirically determined parameters @xcite .
the saturated particle velocity @xmath14 for transport in the aeolian regime can be obtained by iteratively solving eq .
( [ vsaeolian ] ) for @xmath14 and using the expressions for @xmath242 and @xmath250 given by eqs .
( [ eq : v_t ] ) and ( [ f ] ) , respectively .
we calculate the threshold shear velocity @xmath188 by using the following equation , which has been obtained from an analytical model for aeolian sediment transport in equilibrium @xcite , @xmath251 where @xmath252 is given by the following equation @xcite , @xmath253 in the equation above , @xmath254 is an empirically determined parameter @xcite , while @xmath255 , which is the surface roughness of the quiescent sediment bed , is given by the equation @xcite , @xmath256 , \nonumber\end{aligned}\ ] ] where @xmath257 .
we note that the ratio between @xmath188 ( which is the threshold for sustained transport ) and the fluid threshold @xmath258 required to initiate transport in the aeolian regime depends strongly on the environmental conditions .
( [ ut ] ) yields @xmath259 for earth conditions , which is in agreement with measurements @xcite
. however , the ratio @xmath260 under martian conditions can be as small as @xmath261 , as also found from numerical simulations @xcite .
indeed , eq . ( [ ut ] ) , which was obtained from the same theoretical work leading to eq .
( [ vsaeolian ] ) , has been validated by comparing its prediction with outcomes of numerical simulations @xcite under a wide range of fluid - to - sediment density ratio and particle diameter , thereby leading to excellent quantitative agreement ( see fig .
13b of ref .
@xcite ) . in this section ,
we provide expressions for the parameters @xmath90 , @xmath163 and @xmath183 , as well as for the coulomb friction coefficient , @xmath76 , the saturated particle velocity , @xmath14 , and the threshold shear velocity , @xmath188 , for transport in the subaqueous regime .
we remark that we estimate these quantities only in a rough manner , consistent with the large scatter ( factor @xmath210 ) of the experimental data . in this section ,
we reiterate some of the results we obtained in section a2 of the supplementary material of ref .
we can estimate @xmath90 for transport in the subaqueous regime from measurements of the distribution @xmath262 of horizontal velocities @xmath263 in subaqueous sediment transport in equilibrium .
such measurements were undertaken by lajeunesse et al . in experiments of sediment transport under water using particles of average diameter @xmath264 mm and relative shear velocity @xmath265 @xcite . in these experiments , particles
were considered as being transported if they had a velocity larger than a certain cut - off value , @xmath266 @xcite .
the distribution of horizontal velocities for these transported particles was fitted using an exponential distribution , @xmath267},\ ] ] where @xmath268mm@xmath269s . by using this distribution
, we can compute @xmath90 as , @xmath270 lajeunesse et al .
did not report specific values of @xmath266 corresponding to specific measurements @xcite .
instead they mentioned that @xmath266 lies within the range between @xmath271mm@xmath269s and @xmath272mm@xmath269s , depending on the water flow rate .
since @xmath264 mm and @xmath265 ( which are the values reported for the measurement of @xmath262 ) correspond to intermediate values for @xmath273 and @xmath274 investigated in the experiments @xcite , we use the intermediate value @xmath275mm@xmath269s as an approximate estimate for the average cut - off velocity . using this estimate for @xmath266 , eq.([estcv ] )
yields , @xmath276 for transport in the subaqueous regime .
in contrast to the aeolian regime , the suppression of the fluid flow due to the sediment transport in the subaqueous sediment transport is weak @xcite .
the mean fluid speed @xmath3 is thus mainly a function of the shear velocity @xmath95 and the dependence of @xmath3 on @xmath147 and thus on @xmath12 is negligible . by neglecting this dependence , we obtain , @xmath277 which is consequence of eq .
( [ cudef ] ) with @xmath278 . in order to estimate @xmath183 for subaqueous sediment transport
, we use evidence provided by the recent numerical study of durn et al . @xcite . as mentioned before
, these authors simulated the dynamics of both the transported particles and the sediment bed at the single particle scale .
durn et al .
@xcite found that , during flux saturation in subaqueous sediment transport , @xmath11 changes within a time scale which is more than one order of magnitude larger than the time scale in which @xmath0 changes .
this observation can be mathematically expressed as , @xmath279 and thus , @xmath280 eq . ( [ dmvsdv ] ) further implies that , @xmath281 and thus @xmath282 where we used the definition of @xmath183 , which is given by eq .
( [ cmdef ] ) .
hence , we estimate @xmath183 as , @xmath283 however , we note that our model predictions are consistent with experiments even if we assume a coupling of @xmath11 to @xmath12 which is as strong as in the aeolian regime that is , even by assuming @xmath284 and thus increasing @xmath285 by a factor of @xmath286 as compared to the value obtained with @xmath233 @xcite .
this means that the saturation length in the subaqueous regime is not very sensitive to the value of @xmath183 within the range between @xmath287 and @xmath288 ( whereas the latter value corresponds to sediment transport in the aeolian regime ) . in this section
, we reiterate some of the results we obtained in section b of the supplementary material of ref .
as obtained in experiments on subaqueous sediment transport in equilibrium , the average mass flux @xmath13 approximately follows the expression @xcite , @xmath289.\end{aligned}\ ] ] by comparing this equation with eq .
( [ ms ] ) with @xmath290 and @xmath200 ( see section [ equations_for_ls ] ) , we obtain , @xmath291 for sediment transport in the subaqueous regime . it has been verified in a large number of experimental studies @xcite , that the equilibrium particle velocity in the subaqueous regime of transport approximately follows the expression , @xmath292 where @xmath293 is a dimensionless number .
we note that the above expression is consequence of the equation , @xmath294 , where @xmath295 is taken proportional to @xmath95 . in order to obtain @xmath14 for sediment transport in the subaqueous regime using eq .
( [ vsvrs ] ) , we calculate @xmath116 using eq .
( [ vrs2 ] ) and use the value @xmath296 , which we have obtained by comparing the prediction of eq .
( [ vsvrs ] ) with measurements of @xmath14 as a function of @xmath95 from experiments on subaqueous sediment transport in equilibrium @xcite ( see fig . [ aplot ] ) .
, as a function of the dimensionless shear velocity @xmath297 . for the symbols , the average dimensionless particle speed @xmath298 was obtained from measurements @xcite , while we computed @xmath116 using eq .
( [ vrs2 ] ) with @xmath299 .
the black solid line corresponds to the best fit to the experimental data using eq .
( [ vsvrs ] ) , which yields @xmath296 . ] the threshold velocity for sustained sediment transport , @xmath188 , in the subaqueous regime is computed by using the equation , @xmath300 where the threshold shields parameter @xmath301 is obtained through an empirical fit to the shields diagram @xcite . the resulting expression for @xmath301 reads @xcite , @xmath302 where , @xmath303{s\tilde g/\nu^2}$ ] .
in order to understand the morphodynamics of sediment landscapes under water and on planetary surfaces , it is important to understand the behavior of the flux saturation length as a function of the relevant attributes of sediment and fluid . in particular , the size of planetary dunes can serve as a proxy for the saturation length of extraterrestrial dune fields , which can be used to infer the local fluid shear velocity ( @xmath95 ) and average size ( @xmath273 ) of the constituent sediment @xcite .
in fact , the dependence of @xmath2 on @xmath95 has been subject of intense debate in previous theoretical works @xcite .
it is therefore useful to perform in this section a systematic study of the saturation length as a function of these two relevant parameters under different environmental conditions .
[ ls_u ] shows the dependence of @xmath304 on @xmath305 for aeolian sediment transport on earth ( brown solid line ) and mars ( red dashed line ) and for sediment transport under water ( blue dash - dotted line ) computed using eq .
( [ lvfinal2 ] ) for particles with mean diameter @xmath306 . versus dimensionless shear velocity @xmath305 for particles with mean diameter @xmath306 , computed using eq .
( [ lvfinal2 ] ) .
the brown , solid line corresponds to aeolian sediment transport on earth ( @xmath307 , @xmath308 , @xmath309 , @xmath310 ) , the red , dashed line corresponds to aeolian sediment transport on mars ( @xmath311 , @xmath312 , @xmath313 , @xmath314 ) , and the blue , dash - dotted line corresponds to subaqueous sediment transport under water ( @xmath307 , @xmath315 , @xmath309 , @xmath316 ) . moreover , the black , dotted line shows @xmath317 as proposed by refs . @xcite . ] the behavior of @xmath2 with @xmath95 as predicted from eq .
( [ lvfinal2 ] ) is in clear contrast to the scaling relation @xmath318 proposed in previous works @xcite .
this approximate scaling , which includes no dependence of @xmath2 on @xmath95 , was obtained by assuming that the acceleration of transported particles due to fluid drag is the dominant relaxation mechanism , and by neglecting the entrainment of sediment bed particles due to fluid lift as well as the entrainment of sediment bed particles and the deceleration of transported particles resulting from grain - bed collisions . in our more comprehensive model for saturation of sediment flux , however , all these aforementioned relaxation processes are taken into account .
there are two main reasons for the disparity in the behavior of @xmath319 with @xmath95 as observed in our model and in the model of refs .
@xcite .
first , our expression involves a significant dependence of @xmath2 on @xmath305 due to the dependence of @xmath2 on the average particle velocity @xmath14 and the feedback term @xmath175 ( see eq .
( [ lsfluid ] ) ) , both of which are functions of @xmath305 . in particular , for the subaqueous regime , @xmath14 is a strongly increasing function of @xmath305 thus explaining the strong increase of @xmath2 with @xmath305 in this regime .
furthermore , we see in fig .
[ ls_u ] that in our model the dependence of @xmath2 on @xmath95 in the aeolian regime is small , but not negligible as suggested in the model of refs .
indeed , the dependence of @xmath2 on @xmath14 and @xmath175 is a consequence of considering grain - bed collisions and the transport - flow feedback , respectively , for the saturation of the sediment mass flux @xmath0 both neglected in the models of refs .
second , in contrast to the models proposed in these works , our model considers the dependence of the drag coefficient @xmath105 on the particle reynolds number , @xmath320 ( see eq .
[ drag ] ) . as can be seen in fig .
[ ls_u ] , the difference between the particle reynolds numbers on earth ( @xmath321 ) and mars ( @xmath322 ) results in an order of magnitude difference between @xmath304 on these two planetary bodies , which occurs because the normalized particle velocity @xmath323 increases strongly with @xmath324 when @xmath324 is of order unity @xcite . it can also be seen in fig . [ ls_u ] that the saturation length @xmath2 on mars decreases with @xmath220 when @xmath220 is sufficiently close to @xmath325 , even though @xmath14 increases with @xmath95 in this regime .
this surprising behavior is a consequence of the feedback term @xmath175 , which , for mars conditions , decreases with @xmath220 sufficiently close to @xmath325 and thus overcompensates the tendency of @xmath2 to increase with @xmath14 .
in contrast , for earth conditions , the feedback term @xmath175 increases with @xmath220 close to @xmath325 .
this qualitative difference in the change of @xmath175 with @xmath220 between earth and mars conditions can be understood by noting that , @xmath326 for mars conditions with sufficiently small @xmath305 , while for earth conditions ( and for mars conditions with sufficiently large @xmath305 ) the following relation holds , @xmath327 the physical origin for the difference in the behavior of @xmath175 with @xmath95 mentioned above lies in the mechanics of the reduction of the fluid speed due to sediment transport ( see section [ derivation_of_ls ] ) .
the fluid velocity in the transport layer ( @xmath3 ) decreases with the average drag force applied by the fluid onto the transport layer ( @xmath328 ) .
this average drag force in turn is proportional to both the mass density @xmath11 of transported particles and to the acceleration term @xmath329 . recalling that @xmath194 in the aeolian regime , we have that , close to saturation , both @xmath11 and @xmath12 are smaller ( if @xmath330 ) or larger ( if @xmath331 ) than their respective saturated values , @xmath13 and @xmath14 .
hence , if @xmath332 ( @xmath333 ) , it follows that @xmath334 ( @xmath335 ) , which means that the mass density and the acceleration term deviate from their saturated values in opposite directions .
the average drag force applied by the fluid onto the transport layer ( @xmath328 ) thus can be both larger ( under mars conditions for sufficiently small @xmath305 ) or smaller ( under both terrestrial and martian conditions for sufficiently large @xmath305 ) than its saturated value .
consequently , the fluid velocity in the transport layer ( @xmath3 ) can be smaller or larger than its saturated value @xmath295 , depending on whether eq . ( [ diffmarsearth2 ] ) or eq .
( [ diffmarsearth1 ] ) , respectively , is fulfilled .
further , if @xmath336 ( eq . ( [ diffmarsearth1 ] ) ) , then @xmath337 and thus the saturation length @xmath2 decreases in comparison to the situation in which the saturation of feedback is neglected ( @xmath179 ) . in contrast , if @xmath338 ( eq . ( [ diffmarsearth2 ] ) ) , @xmath339 and thus the saturation length @xmath2 increases in comparison to @xmath179 .
the deviation of @xmath175 from @xmath179 becomes stronger with increasing @xmath305 because the effect of the sediment transport on the fluid velocity increases with @xmath305 .
this explains why the feedback term @xmath175 can both increase or decrease with @xmath305 .
[ ls_d ] shows the dependence of @xmath304 on @xmath273 for aeolian sediment transport on earth ( brown solid line ) and mars ( red dashed line ) , and for subaqueous sediment transport ( blue dash - dotted line ) , computed using eq .
( [ lvfinal2 ] ) for different values of the fluid shear velocity , namely , @xmath340 , @xmath341 , and @xmath342 .
it can be seen that the rescaled saturation length @xmath304 displays a complex behavior with @xmath273 .
the dependence of @xmath304 on @xmath273 is controlled by two main factors : the dependence of the drag coefficient and of the shields parameter on the particle reynolds number ( @xmath324 ) , and the dependence of the saturation of the sediment flux on the transport - flow feedback .
the significance of each one of these factors for the dependence of the saturation length on the average grain diameter depends on the transport regime , as we will discuss in the next paragraphs .
_ subaqueous regime of transport _ the change in the transport - flow feedback due to the saturation of the flux is negligible in the subaqueous regime , since in this regime @xmath181 and thus @xmath182 .
this behavior explains why all curves @xmath304 versus @xmath273 in figs .
[ ls_d]a@xmath343c corresponding to transport under water display the same qualitative behavior independent of @xmath95 . in each of these curves
, @xmath304 first increases with @xmath273 , then reaches a maximum , and finally approaches a constant value .
the origin of this behavior is that @xmath2 is proportional to @xmath344 ( cf .
( [ lvfinal2 ] ) ) , and thus its dependence on the grain size is determined by the behavior of @xmath344 with @xmath273 . for sufficiently large particle diameters @xmath273 ( large @xmath324 ) , both the drag coefficient @xmath105 and the shields parameter @xmath301 are approximately independent of @xmath273 , and thus both @xmath345 and @xmath323 are roughly independent of @xmath273 ( see eq .
( [ vrs2 ] ) , as well as eqs .
( [ vsvrs ] ) and ( [ shields ] ) for constant @xmath305 ) , while @xmath346 ( see eq .
( [ cd_fx ] ) ) .
hence , @xmath304 is nearly independent of @xmath273 for sufficiently large particle diameters in the subaqueous regime .
however , for smaller particle diameters @xmath273 ( smaller @xmath324 ) , both @xmath105 and @xmath301 , and thus @xmath345 , @xmath323 and @xmath143 , incorporate a dependence on @xmath273 .
in this regime , @xmath345 increases with @xmath273 , whereas both @xmath323 and @xmath143 decrease with @xmath273 .
the increase of @xmath345 with @xmath273 thereby dominates the behavior of @xmath304 for small particle diameters , while the decrease of @xmath323 and @xmath143 with @xmath273 dominates the behavior of the saturation length with @xmath273 for large particle diameters .
_ aeolian regime under terrestrial conditions _
the same qualitative behavior of @xmath304 with @xmath273 observed in the subaqueous regime occurs for aeolian sediment transport on earth at shear velocities close to the threshold , as can be seen in fig .
two main factors dictate the observed dependence of @xmath304 on @xmath273 for aeolian transport under terrestrial conditions .
first , close to the threshold , the fluid velocity is almost undisturbed by the particle transport , since only a few particles are transported .
hence , the feedback term @xmath175 associated with the aeolian regime of transport under terrestrial conditions when @xmath95 is close to threshold for sustained transport , @xmath325 , is close to unity , as it is in the subaqueous regime .
second , @xmath345 , @xmath323 and @xmath143 behave qualitatively in the same manner with @xmath273 as they do in the subaqueous regime .
therefore , also @xmath304 for aeolian transport under terrestrial conditions with @xmath95 close to @xmath325 depends on @xmath273 in the same manner as it does in the subaqueous regime .
in contrast , for large shear velocities , the qualitative behavior of @xmath304 with @xmath273 observed for aeolian sediment transport on earth is qualitatively different from the one observed in the subaqueous regime ( see figs .
[ ls_d]b@xmath343c ) .
this is because the saturation of the transport - flow feedback for aeolian sediment transport plays a relevant role for large @xmath95 .
in this regime , the approximation @xmath182 is not valid anymore , instead the feedback term @xmath175 follows eq .
( [ kasym ] ) . consequently
, @xmath2 is proportional to @xmath347 and not to @xmath344 .
since @xmath323 decreases with @xmath273 before it reaches an approximately constant value , so does @xmath304 as can be seen in figs .
[ ls_d]b@xmath343c .
_ aeolian regime under martian conditions _ for aeolian transport under martian conditions , @xmath304 shows a qualitative behavior with @xmath273 that is different from the one of aeolian transport under terrestrial conditions .
the origin of this discrepancy is a different qualitative behavior of the normalized average particle velocity @xmath323 , as we explain in the following . in aeolian sediment transport ,
the average velocity of particles impacting and leaving the sediment bed ( @xmath348 ) is nearly independent of properties of the fluid @xcite .
rather , @xmath348 is largely controlled by the characteristics of the sediment bed , for instance by cohesive interparticle forces .
these forces increase in importance with decreasing particle size @xcite .
due to these forces , @xmath349 increases with decreasing @xmath273 ( see eq .
( [ eq : v_t ] ) ) . at the same time , as already mentioned , @xmath345 increases with @xmath273 . both @xmath350 and @xmath345 control the normalized average particle velocity @xmath323 ( see eqs .
( [ vsaeolian ] ) and ( [ eq : v_t ] ) ) . the qualitative difference between aeolian sediment transport on earth and mars
is due to @xmath351 being approximately @xmath352 times larger on earth than on mars .
this difference in the scaling of @xmath351 implies that the decreasing trend of @xmath350 with @xmath273 has a smaller effect on the saturation length for martian conditions than it does for @xmath2 under terrestrial conditions .
indeed , while the value of @xmath323 of aeolian transport under terrestrial conditions decreases with @xmath273 ( due to the decrease of @xmath350 with @xmath273 ) , the martian value of @xmath323 _ increases _ with @xmath273 .
this behavior , which is opposite to the one of terrestrial transport , is due to the increase of @xmath345 with @xmath273 except for small particle diameters for which @xmath345 becomes much smaller than @xmath350 . the complex behavior of the saturation length @xmath2 plotted in figs .
[ ls_u ] and [ ls_d ] suggests that both the entrainment of sediment bed particles by fluid lift and grain - bed collisions , and the momentum change of saltating particles due to drag and grain - bed collisions have a considerable influence on the saturation of the sediment flux .
thus , these relaxation mechanisms of the sediment flux probably can not be neglected as done in previous studies @xcite .
moreover , the relaxation of @xmath3 , which has also been neglected in previous studies , seems to play a significant role in the saturation length of aeolian sediment transport , for which the feedback term @xmath175 ( eq . [ lsfluid ] ) does not vanish since @xmath353 .
in fact , for sufficiently large @xmath220 , the feedback term @xmath175 follows eq .
( [ kasym ] ) , which means that @xmath2 scales with @xmath347 instead of scaling with @xmath344 for small @xmath220 .
hence , our study suggests that the saturation of @xmath3 can not be neglected for aeolian sediment transport .
in conclusion , we have presented a model for flux saturation in sediment transport which , for the first time , accounts for both relevant relaxation processes of sediment flux identified in previous works namely , the saturation of the mass density of transported particles and the relaxation of particle velocities as well as for the different types of sediment entrainment prevailing under different environmental conditions .
furthermore , our model accounts for the saturation transient of the fluid velocity within the transport layer , which is associated with the saturation of the transport - flow feedback inherent to the interaction between the fluid and the particles in transport .
the main outcome of our analytical treatment is a closed - expression for the saturation length of sediment transport , @xmath2 ( eq .
( [ lvfinal2 ] ) ) , which can be used to calculate @xmath2 under different environmental conditions corresponding to both subaqueous and aeolian regimes of sediment transport .
in particular , @xmath2 predicted from our equation is a complex function of the grain diameter , @xmath273 , and of the fluid shear velocity , @xmath95 .
this behavior is in contrast with the scaling of @xmath2 with @xmath354 @xcite , which was obtained from a simplified model that considers only the relaxation of particle velocity and thus neglects the dependence of @xmath2 on @xmath95 observed in experiments @xcite . while the purpose of the present work was to introduce our theoretical model for flux saturation and to present the analytical derivation of our universal equation for the saturation length , in a separate work @xcite
we show that this equation consistently predicts the saturation length in different physical environments .
indeed , our saturation length equation is in good quantitative agreement with direct measurements of @xmath2 in a wind tunnel , as well as with indirect estimates of @xmath2 from the size of subaqueous ripples and dunes on earth , mars and venus . in future studies ,
our equation can be used to predict the scale of dunes under different extraterrestrial environments @xcite or to infer attributes of sediment and flow in planetary dune fields from the minimal size of barchan dunes or from the wavelength of `` elementary '' dunes emerging on dense sand beds . moreover
, our equation can be used to calculate the saturation length in the morphodynamic dune model of ref .
@xcite , which couples a continuum model for sediment transport with an analytical model for the average turbulent fluid shear stress over mildly - sloped topographies @xcite .
our equation can further improve morphodynamic models in hydraulic engineering applications in which the saturation length is usually treated as an adjustable parameter @xcite .
the importance of the saturation length for such models has been debated among engineers @xcite , and it was concluded that even at fluvial scale the influence of sediment transport saturation was significant @xcite . we thus anticipate that the present work can provide substantial contributions to several areas of the geological , planetary and engineering sciences .
we acknowledge support from grants no .
nsfc 41350110226 , no .
nsfc 41376095 , no .
eth-10 - 09 - 2 , no .
nsf ags 1137716 , from the european research council ( erc ) advanced grant no .
319968-flowccs , dfg through the cluster of excellence engineering of advanced materials , and the collaborative research centre sfb814 ( additive manufacturing ) .
we further thank miller mendoza and robert sullivan for fruitful discussions and orencio durn for allowing us to use his numerical model to confirm some of our arguments in section [ derivation ] .
69ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _ _ ( , ) @noop _ _
( , ) @noop _ _ ( , ) @noop _ _ ( , ) link:\doibase 10.1103/physreve.66.031302 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.0800202105 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/14/4/043035 [ * * , ( ) ] in @noop _ _ ( , , ) @noop _ _ ( , ) in @noop _ _ ( ) @noop * * , ( ) link:\doibase 10.1002/esp.1300 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreve.64.031305 [ * * , ( ) ] link:\doibase 10.1016/j.epsl.2006.09.004 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1063/1.2397005 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.111.218002 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) \doibase 10.1061/(asce)0733 - 9429(2008)(134:3)(340 ) [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ , vol .
( ) pp . @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop _ _ ( , ) @noop * * , ( ) link:\doibase 10.1029/2001wr000681 [ * * , ( ) ] link:\doibase 10.1029/2009jf001628 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1098/rsta.1977.0009 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1017/s0022112008005491 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1029/2007jf000774 [ * * , ( ) ] link:\doibase 10.1029/2009jd011702 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.098001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.058001 [ * * , ( ) ] link:\doibase 10.1017/s0022112004009073 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) \doibase 10.1061/(asce)0733 - 9429(1998)124:12(1235 ) [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.074502 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreve.75.011301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.198001 [ * * , ( ) ] link:\doibase 10.1016/j.aeolia.2011.07.006 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1061/(asce)hy.1943 - 7900.0000116
[ * * , ( ) ] link:\doibase 10.1061/(asce)hy.1943 - 7900.0000116 [ * * , ( ) ] \doibase 10.1061/(asce)0733 - 9429(2008)134:4(367 ) [ * * , ( ) ] link:\doibase 10.1061/(asce)hy.1943 - 7900.0000296 [ * * , ( ) ] link:\doibase 10.1680/wama.10.00035 [ * * , ( ) ] | the transport of sediment by a fluid along the surface is responsible for dune formation , dust entrainment and for a rich diversity of patterns on the bottom of oceans , rivers , and planetary surfaces .
most previous models of sediment transport have focused on the equilibrium ( or saturated ) particle flux .
however , the morphodynamics of sediment landscapes emerging due to surface transport of sediment is controlled by situations out - of - equilibrium . in particular , it is controlled by the saturation length characterizing the distance it takes for the particle flux to reach a new equilibrium after a change in flow conditions .
the saturation of mass density of particles entrained into transport and the relaxation of particle and fluid velocities constitute the main relevant relaxation mechanisms leading to saturation of the sediment flux . here
we present a theoretical model for sediment transport which , for the first time , accounts for both these relaxation mechanisms and for the different types of sediment entrainment prevailing under different environmental conditions .
our analytical treatment allows us to derive a closed expression for the saturation length of sediment flux , which is general and can thus be applied under different physical conditions . |
organic devices for electronic applications are usually based on thin film technology @xcite .
this is particularly advantageous , as thin films of organic molecules and polymers can be manufactured easily and cheaply in different ways . over the past few years
, an intense research effort has resulted in rapid improvement of the manufacturing processes , which has allowed the commercialization of products based on organic devices , i.e. organic electronics . in spite of the rapid progress that has taken place on the applied side , the rather low chemical and structural purity of the thin films used in device fabrication
has so far prevented a systematic study of the intrinsic electronic properties of organic semiconductors .
that is because for these films , it is the defects that determine the behavior observed experimentally @xcite . as a consequence , our basic understanding of the electronic properties of organic materials is still limited .
improved chemical and structural purity in organic conductors can be obtained by using single crystals of small conjugated organic molecules .
electronic transport through single crystals of different organic molecules has been investigated in the past by means of time of flight ( tof ) measurements @xcite .
it has been found that the hole mobility is approximately 1 @xmath3/vs at room temperature , increasing up to values in excess of 100 @xmath3/vs with decreasing temperature .
since these observations have been reported only in the highest purity crystals , it is believed that this is the _ intrinsic _ behavior of charge carrier mobility in organic conductors .
so far however , this intrinsic behavior has never been observed in conventional dc transport measurements .
, is transported through the system by the carrier gas ( indicated by the arrows ) and recrystallizes at temperature @xmath4 .
heavy impurities ( with a vapor pressure lower that that of tetracene ) remain at the position of the source material .
light impurities ( with a vapor pressure higher than that of tetracene ) condense at a lower temperature @xmath5 , i.e. at a different position from where the crystals grow .
therefore , the crystal growth process also results in the purification of the material .
( b ) result after first regrowth of as - purchased tetracene .
purified tetracene crystals are visible in the middle ; the dark residue present where the source material initially was and the light ( yellow ) material visible on the right are due impurities .
( c ) at the end of the second regrowth no dark residue is present at the position of the source material , which demonstrate the purifying effect of the growth process .
[ growthpictures],width=321 ] in this paper we report an experimental study of dc transport through tetracene single crystal and show that our results exhibit some of the features expected for the intrinsic behavior of organic conductors .
our investigations are based on the study of the current - voltage ( @xmath0-@xmath1 ) measurements in the space charge limited current regime and on their comparison to tof measurements performed on identically grown crystals . as we will show , we find overall agreement between the results obtained with the two different methods .
in particular , both @xmath0-@xmath1 and tof measurements show that the room - temperature effective hole - mobility reaches values close to @xmath2 @xmath3/vs .
both measurement techniques also show that , within a range of temperatures , the mobility increases with decreasing temperature .
for the best samples probed by @xmath0-@xmath1 measurements this range extends down to approximately 200 k , below which a structural phase transition known to occur in tetracene causes a sudden drop of the mobility .
the experimental results further allow the characterization of different aspects of the tetracene crystals .
in particular , we observe the effect of both deep and shallow traps . for the former ,
the measurement of the @xmath0-@xmath1 characteristics give us an upper bound on their bulk density ( @xmath6 @xmath7 ) and an estimate of their depth ( @xmath8 mev relative to the edge of the valence band ) .
the concentration of shallow traps is substantially larger and only a very rough estimate can be obtained from tof experiments .
the paper is organized as follows .
we first describe the most important aspects of the tetracene crystals growth and of the sample preparation ( section [ crystalgrowth ] ) .
the behavior of the measured @xmath0-@xmath1 curves is presented in section [ dctransport ] . in this section
we also discuss the basic aspects of the theoretical concepts necessary to interpret the experimental data .
section [ tof ] is devoted to tof experiments .
finally , in section [ discussion ] we summarize and compare the outcome of tof and dc transport measurements and we present our conclusions .
single tetracene crystals are grown by means of physical - vapor deposition in a temperature gradient ( see fig .
[ growthpictures ] ) , in the presence of a stream of carrier gas , using a set - up similar to that described in reference @xcite . both ar and h@xmath9 were used .
all experiments discussed in this paper are performed on ar grown crystals on which we have obtained the highest values of charge carrier mobility .
crystal growth is performed in the dark to minimize possible photo - activated chemical reaction of tetracene with remnant o@xmath9 . these photo - induced reaction with o@xmath9
are known to occur for most polyacenes @xcite and result in chemical impurities that can act as traps for charge carriers .
-@xmath1 measurement perpendicular to the _ ab _ plane of a tetracene single - crystal , with a thickness @xmath10 m and a mobility @xmath11 = 0.59 @xmath3/vs .
the inset shows a similar measurement on a different crystal ( @xmath12 m , @xmath11 = 0.014 @xmath3/vs ) , in which a crossing over into an approximately quadratic dependence on voltage is visible at high voltage . in both cases ,
a very steep current increase occur around of just above @xmath13 that we attribute to filling of deep traps .
we observed a steep increase in current in most samples studied .
[ typicalsclc],width=321 ] the source material for the first crystal growth is @xmath14 pure tetracene purchased from sigma - aldrich .
crystals grown from as - purchased tetracene are used as source material for a subsequent growth process . for the first and second growth steps
the results of the growth processes are shown in fig .
[ growthpictures ] .
note the large residues of impurity molecules clearly visible after the first process ( fig .
[ growthpictures]b ) but not after the second ( fig . [ growthpictures]c ) , which directly demonstrate the usefulness of the second regrowth process .
tetracene crystals grown by physical - vapor deposition are large platelets , with surfaces parallel to the _ ab _ plane .
typical dimensions range from few square millimeters to @xmath15 @xmath3 or larger . for crystals grown by letting the growth process proceed overnight , the typical crystal thickness ranges between @xmath16 m and @xmath17 m . we have performed x - ray structural study on a few of our thickest crystals and found a structure consistent with known literature data @xcite .
multiply regrown crystals are inspected under an optical microscope with polarized light . as it is typical for many organic conjugated molecules ,
tetracene crystals are birefringent .
this allows us to select single crystals to be used for transport experiments by choosing those samples that become uniformly dark when changing the orientation of cross - polarizers @xcite .
all @xmath0-@xmath1 and tof measurements discussed in this paper have been performed in the direction perpendicular to the crystal _ ab _ plane , using electrical contacts on the two opposite faces of crystals .
electrodes for @xmath0-@xmath1 measurements are fabricated by connecting gold wires to the crystal surface using a two - component , solvent - free silver epoxy , so that the epoxy is in direct contact with the crystal .
the contact area is measured under the microscope and it is typically of the order of 0.1 mm@xmath18 .
we use silver epoxy cw2400 ( circuitworks ) , which hardens at room temperature in a few hours .
this contact fabrication method is very quick and it was chosen because it allows the investigation of many samples in relatively short time .
other types of contacts were tested as well , i.e. metal evaporated contacts and colloidal graphite paint contacts , but they did not result in any improvement of the observed electrical properties . for tof measurements we used silver electrodes prepared by thermal evaporation through a shadow mask .
in this section we discuss the results obtained by studying the @xmath0-@xmath1 characteristics of approximately 100 tetracene single crystals .
we have found that the measured @xmath0-@xmath1 curves exhibit large sample - to - sample deviations so that particular care has to be taken in the interpretation of the experimental data .
for this reason , we first discuss how the charge carrier mobility can be estimated using concepts of space charge limited current theory of general validity , i.e. not sensitive to the detailed behavior of our samples .
after having presented the experimental results in terms of the concept previously introduced , we discuss the role of deep traps present in the bulk of the crystals and at the metal / organic contact interface .
we argue that the latter provide the most likely explanation of the large sample to sample variation observed in the measured @xmath0-@xmath1 characteristics .
since the band - gap of tetracene is approximately @xmath19 ev @xcite , high - purity tetracene crystals containing a negligible amount of dopants essentially behave as insulators .
it is still possible to pass a current through tetracene crystals by applying a sufficiently large voltage , which acts both to transfer charge from the electrodes into the crystal and to accelerate that charge .
when the charge injected from the contacts is larger than the charge present in the material in equilibrium the @xmath0-@xmath1 characteristics become non - linear and transport is said to occur in the space charge limited regime @xcite . for materials in which current is carried by only one carrier type ( holes in our tetracene crystals ;
see section [ tofmeas ] ) , there exists an upper limit to the current that can be carried in the space charge limited regime .
this is due to electrostatics that , at any given voltage @xmath1 , fixes the maximum amount of charge that can be injected into the material . for the geometry used in our experiments , the resulting maximum current that can flow in the space charge limited regime in the presence of an applied voltage @xmath1 is @xcite : @xmath20 with @xmath21 and @xmath22 respectively electrode area and separation and @xmath23 relative dielectric constant of the material ( @xmath24 for tetracene ) .
this upper limit is _ intrinsic _ and it is not sensitive to any of the specific sample details that determine the shape of the @xmath0-@xmath1 curves . for any applied voltage ,
contact effects , defects or traps can only reduce the current below the value given by equation [ sclc ] .
we use these considerations to obtain experimentally a lower limit @xmath11 for the mobility of charge carrier , by measuring the current @xmath0 induced by a voltage @xmath1 and by `` inverting '' equation [ sclc ] .
we obtain : @xmath25 if @xmath11 is very low - e.g. , at room temperature , much lower than the intrinsic value 1 @xmath3/vs typical of organic semiconductors - this approach does not provide any useful information .
however , if the value of @xmath11 is close to 1 @xmath3/vs , this analysis indicates that the quality of the crystal is high ( since the intrinsic mobility @xmath26 ) .
all the measurements of @xmath0-@xmath1 characteristics of tetracene crystals discussed in this paper have been performed in vacuum ( @xmath27 mbar ) and in the dark , in a two - terminal configuration .
we have used a keithley 237 source - measure unit that permits to apply up to 1100 v across our samples and to measure currents as small as 10 fa .
measurements at lower temperature were performed in the vacuum chamber of a flow cryostat .
approximately 100 samples have been measured at room temperature .
temperature dependent measurements have been performed on most samples in which a high value for @xmath11 ( 0.1 @xmath3/vs or better ) has been found and on few of the others . calculated from dc @xmath0-@xmath1 measurements performed on approximately 100 tetracene single - crystals .
[ sclcmobilitiesoverview],width=321 ] the precise shape of the @xmath0-@xmath1 characteristics measured on different samples exhibit large deviations , whose possible origin is discussed in the next section . here , we focus on some important common features observed in many of the samples investigated ( @xmath28 ) .
in particular , we often observe that in the lower voltage range ( @xmath29 v , depending on the sample ) the current increases with voltage is approximately quadratic . at higher voltage , the current increases by many decades ( typically six to eight , depending on the sample ) for a one - decade increase in voltage ( fig .
[ typicalsclc ] ) . in most cases samples fail as the voltage is increased in this part of the @xmath0-@xmath1 curve , either because too much power is dissipated through the crystals ( samples with high @xmath11 ) or because the electrical contacts detach from the crystal ( samples with low @xmath11 ) . in a few cases , however ,
we have observed that the rapid current increase terminates by crossing over into an approximately quadratic dependence on voltage ( fig .
[ typicalsclc ] , inset ) . , measured for several tetracene single - crystals .
note the abrupt drop in mobility occurring at different temperatures below @xmath30 k , originating from a known structural phase transition @xcite .
[ sclcmobvstemp],width=321 ] the value of the current measured at the maximum applied voltage is used to calculate @xmath11 from equation [ sclcinv ] .
the calculation requires the knowledge of the crystal thickness , which , in this geometry , corresponds to the electrode separation @xmath22 .
the measurement of @xmath22 is done by inspecting the crystals under an optical microscope .
the uncertainty of @xmath22 ( typically @xmath31 to @xmath32 ) is rather large and due to the difficulty of the measurement and to the opposite crystals surfaces not being parallel to each other . in calculating @xmath11
we have used lower estimates of @xmath22 , in order to be sure not to over - estimate the crystal mobility . the histogram shown in fig .
[ sclcmobilitiesoverview ] provides an overview of the values of @xmath11 obtained from all measured samples .
the spread in the calculated values of @xmath11 is large , as a consequence of the large deviations observed in the measured @xmath0-@xmath1 characteristics . in what follows we concentrate on those samples for which the lower limit to the mobility is @xmath33 @xmath3/vs . for most high - quality samples that did not fail during the room - temperature measurements we have performed @xmath0-@xmath1 measurements at different temperatures .
we reproducibly find that @xmath11 increases upon lowering temperature in all the samples in which @xmath33 @xmath3/vs at room temperature ( fig .
[ sclcmobvstemp ] ) .
the same behavior has also been observed in a couple of samples in which @xmath34 @xmath3/vs , although normally samples for which @xmath35 @xmath3/vs at room temperature exhibit a decrease in current as the temperature is lowered .
we conclude that samples in which the room - temperature mobility is sufficiently close to the value of 1 @xmath3/vs exhibit the behavior expected for high - quality organic semiconductors , i.e. _ an increasing mobility with lowering temperature_. this is the first time that this behavior is reported in simple two - terminal dc @xmath0-@xmath1 characteristics . in all the high - mobility samples measured
we observe that below @xmath36 , @xmath11 starts to decrease when the temperature is decrease further . in most samples ,
the change in the temperature dependence of @xmath11 is very sharp ( fig .
[ sclcmobvstemp ] ) .
this suggests that the origin of this change is a structural phase transition , which is known to occur in tetracene in this temperature range . past studies of this transition @xcite
have shown that the precise transition temperature depends on details , such as the stress induced by the adhesion between a crystal and the substrate on which the crystal is mounted .
these studies have also shown that the phase transition does not occur uniformly , with two different crystalline structures coexisting in different parts of a same crystal in a large interval of temperatures below the transition .
the coexistence of different crystal phases is detrimental for transport , since it introduces grain boundary junctions and regions with different bandwidth that can trap large amounts of charge carriers .
this explains the observed temperature dependence of the mobility .
the observation of the effect of this structural phase transition on the transport properties of tetracene provides one additional indication that the crystal quality is high and that we are probing intrinsic effects in the material .
the interpretation of experimental data in terms of @xmath11 is of general validity and it does not require any assumption regarding the sample characteristics .
additional information can be extracted from the measurements if one considers the behavior of the measured @xmath0-@xmath1 curves in more detail .
here we consider the effect of deep traps present in the bulk of the tetracene crystals and at their surface . in general
, deep traps suppress current flow by localizing charge carriers . in the space charge limited transport regime , it can be easily shown @xcite that when the applied voltage @xmath1 is approximately equal to @xmath37 the charge injected by the contacts is sufficient to fill all the traps and transport occurs in the so - called trap - filled limit . at this point (
i.e. , with increasing @xmath1 from below to above @xmath38 ) the measured current exhibits a large sudden increase given by : @xmath39 in this expression , @xmath40 is the number of states in the valence band which we take to be of the order of one state per molecule @xcite . essentially all samples exhibit a large , steep increase in current around a given ( sample dependent ) voltage ( fig . [ typicalsclc ] ) , which we interpret as due to the transition to the trap filled limit . using equation [ vtfl ]
we obtain @xmath41 @xmath7 .
we find that different samples all give comparable values .
introducing this value for @xmath42 in equation [ trapfillingjump ] we then find , taking the magnitude of current increase measured on samples with the highest value of @xmath11 , @xmath8 mev .
the estimate of @xmath42 is based on the assumption , not usually emphasized in literature , that the deep traps are uniformly distributed throughout the entire crystal bulk . in actual samples , due to the contact preparation process , it is likely that more traps are present at the crystal surface under the electrodes .
a small amount of traps located close to the surface can have a large effect in suppressing current flow .
this is because charges trapped at the surface can substantially affect the electrostatic profile in the bulk of the crystal , which determines the current flow in the space charge limited current regime . to make this point more explicit ,
consider a @xmath43 m thick crystal in which no traps are present apart from in the first monolayer of molecules close to the surface .
suppose that , in this monolayer , one deep trap per every 1000 molecules is present .
this will result in a very large current suppression , as populating these traps results in an electric field through the crystal which corresponds , in the case considered , to approximately 1000 volts applied between the electrodes .
for an @xmath0-@xmath1 measurement this would imply that , as the voltage across the electrodes is increased , the surface traps are initially filled and no current flows until more than 1000 volts are applied . for comparison , for a @xmath43 m thick crystal with no surface traps and
a bulk density of traps of @xmath44 @xmath7 transport already occurs in the trap free limit when 200 v are applied across it . for this reason ,
our estimate of @xmath42 is a higher limit to the bulk density of deep traps .
the strong sensitivity of the @xmath0-@xmath1 curves to traps located at the metal / organic interface makes these traps a logical explanation for the large sample to sample variation observed in the experiments .
evidence for the relevance of contact effects is provided by the rather good sample - to - sample reproducibility observed in tof measurements ( see section [ tofmeas ] ) as compared to dc transport measurements .
the time of flight ( tof ) spectroscopy is based on two fundamental steps .
first , by photo - excitation electron hole pairs are generated near the crystal surface .
second , in the applied electric field the charge carriers move to the electrodes and the corresponding displacement current is measured @xcite .
therefore , studies on the transport behavior are free of contact effects and the technique is selective on the respective type of charge carrier by choice of the polarity of the external voltage . for the tof measurements
the tetracene crystals are covered on both sides by a thin layer of ag thermally evaporated and , afterwards , mounted on a cu support acting as back - electrode as well as thermal contact for the temperature dependent studies .
the @xmath45 nm thick ag contacts showing almost bulk conductivity but are still sufficiently transparent for photo - induced generation of charge carriers at the front electrode . as light source for the charge carrier generation , a nitrogen laser in single shot mode (
@xmath46 = 337 nm , pulse width 0.76 ns ) is used .
the suitability of the used wavelength for photo - excitation has been proven by uv - vis absorption measurements on the crystals .
in addition , from the tof pulse shape and the absorption spectra we can conclude that charge carrier generation takes place in the topmost fraction of the crystals mainly , i.e. the depth of charge carrier generation ( several @xmath47 m ) can be neglected with respect to the thickness of the crystals ( @xmath48 m ) . c. the applied voltage is + 500 v. the arrows point to the transit times .
[ stuttgart1],width=321 ] the displacement current is measured as a voltage drop across a resistor connected in parallel to the crystal .
the resistor is chosen in such a way that the time constant of the rc - part is much smaller than the transit time @xmath49 .
tof studies in the range from room temperature up to 450 k were carried out in a heating stage at ambient pressure , the upper temperature limit caused by sublimation of tetracene at around 450 k @xcite .
temperature dependent tof measurements have been performed on three different single crystals giving essentially identical results .
the room - temperature mobility values obtained from the different crystals were very close to each other ( ranging from @xmath50 @xmath3/vs to @xmath51 @xmath3/vs ) and their temperature dependence was qualitatively identical .
this is in net contrast with the large sample - to - sample variations observed in dc transport measurements . from the outcome of the tof experiments we conclude that the quality of the grown crystals is rather reproducible and that the sample to sample variations observed in the dc @xmath0-@xmath1 originates from irreproducibility in the contact preparation
this is consistent with the fact that tof experiments are not very sensitive to the contact quality whereas dc @xmath0-@xmath1 measurements are .
representative tof pulses for positive charge carriers measured at various temperatures in one of the three tetracene single crystals are shown in fig .
[ stuttgart1 ] , from which the transit time @xmath49 can be easily extracted .
in contrast to these well - defined tof pulses measured for hole transport , only dispersive transport is observed for electrons throughout the measured temperature range .
this indicates strong trapping for electrons , which is why in section [ dctransport ] we have used single carrier space charge limited current theory to interpret the behavior of the measured @xmath0-@xmath1 curves . assuming a constant electric field _
e _ across the crystal , the mobility of the holes is related to the transit time and to the crystal thickness @xmath22 by : @xmath52 for the values of electric field used in our studies @xmath53 depends linearly on @xmath54 ( see fig . [ stuttgart2 ] ) so that @xmath47 does not depend on electric field .
all the mobility values discussed in this paper have been estimated from this ohmic regime only .
values gives a measure of the uncertainty in the extracted mobility .
[ stuttgart2],width=321 ] in the absence of traps equation [ tofmobility ] represents the intrinsic mobility @xmath47 of the organic material , which typically varies as an inverse power of temperature , i.e. @xmath55 ( @xmath56 depending on the specific organic molecules @xcite ) .
if shallow traps are present , however , the measured mobility is just an `` effective '' mobility @xmath57 related to the @xmath47 by @xcite : @xmath58 } \label{mueff}\ ] ] where @xmath59 is the density ratio between shallow traps and organic molecules in the crystal and it is assumed that all the shallow traps have the same energy depth @xmath60 relative to the valence band .
a plot of the measured @xmath57 vs. @xmath61 is shown in fig . [ stuttgart3 ] .
the saturation with lowering temperature is a characteristic signature of shallow traps .
the effect of shallow traps is not only visible in the temperature dependence of the mobility , i.e. of the transit time of tof pulses , but also in their shape .
specifically , fig .
[ stuttgart1 ] shows that the signal intensity increases with increasing temperature and that the pulse becomes more rectangular .
both effects are due to reduced trapping at elevated temperature , which make the sample behave more closely to the ideal trap - free condition .
using an exponent @xmath62 , a mobility of 1.4 @xmath3/vs , a trap energy of 0.13 ev and a density ratio between shallow traps and tetracene molecules of @xmath63 .
[ stuttgart3],width=321 ] we have attempted to use equation [ mueff ] to fit the measured temperature dependence of @xmath57 and to determine the values of @xmath64 and @xmath60 . however , the limited temperature range which is experimentally accessible , the unknown intrinsic room - temperature mobility of tetracene and of the value @xmath65 determining its temperature dependence make it impossible to determine trap density and concentration precisely .
even when we set @xmath47(300 k ) = 1 @xmath3/vs and @xmath62 , we find that different combinations of @xmath64 and @xmath60 produce a satisfactory fit of our data . from this analysis
we can however roughly estimate that @xmath66 @xmath7 and @xmath67 mev .
a more precise determination of these parameters would require to extend the temperature dependent measurements over a wider temperature range .
in practice , however , the temperature range is limited by sublimation of tetracene ( @xmath68 k ) on the high end and by the blurring of the tof pulse on the low end ( @xmath69 k ) .
this is the first time that systematic dc transport measurements and tof experiments have been performed on identically grown _
single crystals _ of organic molecules and it is interesting to find overall agreement in the comparison of the results obtained with the two techniques . specifically , from both measurement techniques we conclude that the room temperature hole mobility is close to 1 @xmath3/vs and that its temperature dependence is non - monotonic . for tof measurements on tetracene , this behavior had been observed previously @xcite . on the contrary ,
an increase in mobility with lowering temperature observed by measuring the dc @xmath0-@xmath1 curves in the space charge limited transport regime had not been reported previously neither for tetracene nor , to the best of our knowledge , for any organic un - doped semiconductor .
this observation , together with the observed effect of the structural phase transition on the hole mobility , indicate that signatures of the intrinsic electronic properties are visible in the @xmath0-@xmath1 measurements , which demonstrates the high quality of the crystals . in spite of the crystal quality , the effect of imperfections is still clearly visible . in particular , @xmath0-@xmath1 and tof measurements
provide complementary information about the presence of deep and shallow traps .
@xmath0-@xmath1 measurements allow us to infer an upper limit for the bulk density of deep traps and their activation energy ( see section [ traps ] ) . for shallow traps
the observation of a maximum in mobility in tof measurements around room temperature indicates that the typical activation energy is @xmath70 mev , but a more precise value as well as a reliable estimate of the density can not be presently obtained .
we believe that these shallow traps originate from local deformations of the crystals , such as those due to mechanical stress or to electrically - inactive chemical impurities as well as from chemical impurities interacting only weakly with the surrounding host molecules . within a conventional band picture
it is easy to see that these deformations would have an important effect , as they can change locally the band gap of tetracene and form spatially localized `` pockets '' of holes .
this mechanism could account for a fairly large density of shallow traps and for a trapping energy of @xmath70 mev @xcite , which corresponds to only a few percent change in the 3 ev tetracene gap .
not only the similarities but also the differences between @xmath0-@xmath1 and tof measurements provide useful information . particularly noticeable
is the reproducibility of tof measurements in contrast to the large sample to sample deviations observed in the @xmath0-@xmath1 characteristics .
three out of three crystals studied by means of tof gave @xmath71 @xmath3/vs , whereas out of approximately 100 sample measured only 5 to 10% gave a mobility larger than 0.1 @xmath3/vs .
the reproducibility obtained in tof measurements indicate that different tetracene crystals grown in our set - up exhibit only minor differences in their properties and that these differences can not account for the large spread of @xmath0-@xmath1 characteristics observed experimentally .
we conclude that the large sample to sample variations observed in the measurement of @xmath0-@xmath1 characteristics mainly originate from the quality of the electrical contacts . as mentioned above , this is critically important for @xmath0-@xmath1 measurements , but not for tof measurements . since the large values of @xmath11 obtained in the best samples indicate that it is possible to fabricate `` high - quality '' contacts using silver epoxy , we infer that the effects determining the contact quality in our samples are mainly of _ extrinsic _ nature .
our estimates suggest that deep - traps present under the contacts at the crystal surface ( introduced during the contact fabrication ) play an important role .
this is because even a very small surface density of these traps can substantially perturb the electrostatic profile in the crystal bulk that determines the current flow in the space charge limited transport regime .
we gratefully acknowledge n. karl for his invaluable help and support .
this work was financially supported by fom and by dfg - schwerpunkt ( project no .
ka 427/8 ) .
the work of afm is part of the nwo vernieuwingsimpuls 2000 program .
gelinck , t.c.t .
geuns , and d.m
. de leeuw , appl .
77 , 1487 ( 2000 ) c.d .
dimitrakopoulos , and p.r.l .
malenfant , adv . mater . * 14 * , 99 ( 2002 ) i.h .
campbell , and d.l .
smith , solid state phys . * 55*,1 ( 2001 ) w. warta , and n. karl , phys .
b * 32 * , 1172 ( 1985 ) n. karl , k .- h . kraft , j. marktanner , m. mnch , f. schatz , r. stehle , and h .-
uhde , j. vac .
a * 17 * , 2318 ( 1999 ) r.a .
laudise , ch .
kloc , p.g .
simpkins , and t. siegrist , j. crystal growth * 187 * , 449 ( 1998 ) r. dabestani , m. nelson , and m.e .
sigman * 64 * , 80 ( 1996 ) j. monteath robertson , v.c .
sinclair , and j. trotter , acta cryst .
* 14 * , 697 ( 1961 ) d. holmes , s. kumaraswamy , a.j .
matzger , and k.p.c .
vollhardt , chem .
j. * 5 * , 3399 ( 1999 ) j. vrijmoeth , r.w .
stok , r. veldman , w.a .
schoonveld , and t.m .
klapwijk , j. appl . phys . * 83 * , 3816 ( 1998 ) r. farchioni , and g. grosso , _ organic electronic materials _ ( springer - verlag , berlin , 2001 ) m.a . lampert , and p. mark , _ current injection in solids_(academic press inc . , new york , 1970 ) u. sondermann , a. kutoglu , h. bssler , j. phys
. chem . * 89 * , 1735 ( 1985 ) j.d .
cox , g. pilcher : thermochemistry of organic and organometallic compounds , academic press , new york , 1970 j. berrehar , p. dellanoy , and m. schott , phys .
( b ) * 77 * , k119 ( 1976 ) e.a .
silinsh , and v. pek , _ organic molecular crystals _
( aip press , woodbury , 1994 ) | we report on a systematic study of electronic transport in tetracene single crystals by means of space charge limited current spectroscopy and time of flight measurements .
both @xmath0-@xmath1 and time of flight measurements show that the room - temperature effective hole - mobility reaches values close to @xmath2 @xmath3/vs and show that , within a range of temperatures , the mobility increases with decreasing temperature .
the experimental results further allow the characterization of different aspects of the tetracene crystals .
in particular , the effects of both deep and shallow traps are clearly visible and can be used to estimate their densities and characteristic energies .
the results presented in this paper show that the combination of @xmath0-@xmath1 measurements and time of flight spectroscopy is very effective in characterizing several different aspects of electronic transport through organic crystals . |
the death of massive stars and the processes that lead to the formation of the compact remnants is a forefront area in stellar astrophysics .
recent advances in modeling suggest that great diversity can be expected .
indeed , observationally we have already witnessed a large diversity in the neutron star remnants : radio pulsars , axps , sgrs , and the central source in casa .
we know relatively little about the formation of black holes , static or rotating . the compact object form following the collapse of the progenitor core . the energy of the resulting explosion could be supplemented or even dominated by the energy released from the compact object ( e.g. a rapidly rotating magnetar or an accreting black hole ) .
such `` engines '' can give rise to asymmetrical explosions @xcite .
regardless of the source of energy , a fraction of the total energy , @xmath2 , is coupled to the debris or ejecta ( mass @xmath3 ) and it is these two gross parameters which determine the appearance and evolution of the resulting explosion .
equivalently one may consider @xmath2 and the mean initial speed of ejecta , @xmath4 , or the lorentz factor , @xmath5^{-1/2}$ ] , where @xmath6 .
supernovae ( sne ) and @xmath0-ray bursts ( grbs ) , are distinguished by their ejecta velocities . in the former @xmath7 km s@xmath8
as inferred from optical absorption features ( e.g. @xcite ) , while for the latter @xmath9 , inferred from the non - thermal prompt emission @xcite , respectively . the large difference in initial velocity arises from significantly different ejecta masses : @xmath10 m@xmath11 in sne compared to @xmath12 m@xmath11 in grbs . in the conventional interpretation ,
@xmath3 for sne is large because @xmath2 is primarily derived from the ( essentially ) symmetrical collapse of the core and the energy thus couples to all the mass left after the formation of the compact object .
mysteriously , @xmath2 clusters around 1 foe ( foe is @xmath13 erg ) in most sne , a mere @xmath14 of the energy released in the gravitational collapse of the core . whereas the initial ejecta speed is solely determined by @xmath2 and @xmath3 , a fraction of the ejecta is accelerated to higher velocities as the blast wave races down the density gradient of the stellar enveloped ( e.g. @xcite ) . for the wind- or binary - stripped
( e.g. @xcite ) helium and carbon progenitors of type ib and ic sne , both factors ( a smaller core mass and a steep density gradient ) conspire to produce ejecta at velocities as high as @xmath15 .
however , only @xmath16 is carried by these ejecta @xcite .
in contrast , high velocity ejecta is neither expected nor observed in type ii sne with their massive stellar envelopes .
grb models , on the other hand , appeal to a stellar mass black hole remnant , which accretes matter on many dynamical timescales and powers relativistic jets ( the so - called collapsar model ; @xcite ) ; highly magnetized neutron stars have also been proposed ( e.g. @xcite ) .
observationally , this model is supported by the association of some grbs with sn explosions ( e.g. @xcite ) . in addition , the complex temporal profiles and long duration of grbs are interpreted in terms of an engine that is relatively long lived ( i.e. not a singular explosion ) .
the high lorentz factors , a high degree of collimation with opening angles of a few degrees @xcite , and episodes of energy addition presumably from shells of ejecta with varying lorentz factors , further distinguish grbs from type ib / c sne .
we now recognize that engine - driven events grbs and the recently discovered x - ray flashes @xcite in fact have a wide dispersion in their ultra - relativistic output as manifested by their beaming - corrected @xmath0-ray energies @xcite and x - ray luminosities @xcite .
however , these cosmological explosions appear to have a nearly constant total explosive yield when taking into account the energy in mildly relativistic ejecta @xcite .
the unusual and nearby ( @xmath17 mpc ) sn1998bw shares some of the unique attributes expected of grbs .
this type ic sn was found to be coincident in time and position with grb980425 @xcite , an event with a single smooth profile .
the inferred isotropic energy in @xmath0-rays of grb980425 was only @xmath18 erg @xcite , three to six orders of magnitude fainter than typical grbs .
more importantly , sn1998bw exhibited unusually bright radio emission indicating about @xmath19 erg of mildly relativistic ejecta @xcite .
equally significant , the radio emission indicated a clear episode of energy addition @xcite .
none of these features
@xmath0-rays , significant energy with @xmath20 , and episodes of energy addition have been seen in any other nearby sn .
thus , the empirical data strongly favor an engine origin for sn1998bw . in the optical
, sn1998bw also appears to be extreme : velocity widths approaching 60,000 km s@xmath8 were seen at early time @xcite and the inferred explosion energy may be above normal values , with estimates ranging from 2 to 50 foe @xcite .
the inference of an engine in sn1998bw raises two scenarios for its origin and relation to grbs .
grb980425 may have been a typical burst but viewed well away from the jet axis ( hereafter , the off - axis model ) , thereby resulting in apparently weak @xmath0-ray emission despite the great proximity .
alternatively , sn1998bw represents a different class of sne .
if so , collapsars can produce very diverse explosions .
a powerful discriminant between these two scenarios is the expected rate of sn1998bw - like events . in the off - axis model , the fraction of type ib / c sne that are powered by a central engine
is linked to the mean beaming factor of grbs ( e.g. @xcite ) .
recently , @xcite presented the distribution of jet opening angles for a sample of 15 grbs , and found that the mean beaming factor is @xmath21 ; here @xmath22 $ ] is the beaming fraction , and @xmath23 is the collimation angle . with an estimated local grb rate of @xmath24 gpc@xmath25 yr@xmath8 @xcite
compared to a type ib / c sn rate of @xmath26 gpc@xmath25 yr@xmath8 @xcite , we expect that @xmath27 of type ib / c sne will be similar to sn1998bw .
on the other hand , if sn1998bw is not an off - axis cosmological burst , then the rate of similar events has to be assessed independent of the grb rate .
an upper limit can be obtained by assuming that all type ib / c sne are engine driven highly asymmetric explosions with sn1998bw having the most favorable geometry . in this context , @xcite has argued that of the @xmath28 long - duration batse bursts , about 90 events possess similar high - energy attributes as that of grb980425 .
this sub - sample may be concentrated along the super - galactic plane . if this sub - sample is accepted as distinct from the cosmological bursts then @xmath29 of type ib / c sne within 100 mpc
are expected to be events like sn1998bw . here , we report a comprehensive program of radio observations of nearby type ib / c sne .
we began this program in 1999 ( motivated by sn1998bw ) and observed most reported type ib / c sne with the very large array .
our basic hypothesis is that ( mildly ) relativistic ejecta are best probed by radio observations , as was demonstrated in the case of sn1998bw . furthermore , radio observations of type ib / c sne allow us to directly compare these objects to the radio afterglows of cosmological grbs .
thus , we can empirically ( direct comparison of radio luminosity distributions ) and quantitatively ( calorimetry via radio observations ) investigate the link , or lack thereof between type ib / c sne and cosmological grbs .
as alluded to above , we did not investigate type ii sne since the extended envelopes and dense circumstellar media of their progenitors are reasonably expected to mask the activity of a putative engine and thus suppress the presence of mildly relativistic ejecta to which we are most sensitive .
the organization of the paper is as follows . in [ sec : obs ] we present the details of the observations .
the results are summarized in
[ sec : stat ] , where we investigate the broad radio properties ( [ sec : lums ] ) , expansion velocities ( [ sec : vels ] ) , and energies in high velocity ejecta ( [ sec : energy ] ) .
we further provide a comparison to the radio afterglows of grbs in [ sec : comp ] and draw conclusions in [ sec : conc ] .
table [ tab : data ] summarizes the very large array ( vla ) observations of type ib / c sne starting in late 1999 and up to the end of 2002 . we observed a total of 33 sne out of 51 identified spectroscopically during the same period .
the observed targets were determined solely by the availability of observing time and optical selection criteria ; we did not employ any additional selection criteria . in all observations we used the standard continuum mode with @xmath30 mhz bands , centered on 1.43 , 4.86 , or 8.46 ghz .
we used the sources 3c48 ( j0137 + 331 ) , 3c147 ( j0542 + 498 ) , and 3c286
( j1331 + 305 ) for flux calibration , and calibrator sources within @xmath31 of the sne to monitor the phase .
the data were reduced and analyzed using the astronomical image processing system @xcite .
in this section we investigate the ejecta properties and diversity of the sample .
results for individual sne are given in the appendix . in figure
[ fig : lcs ] we plot the radio luminosities and upper limits for type ib / c sne observed in this survey and in the past ( sn1983n : @xcite ; sn1984l : @xcite ; sn1990b : @xcite ; sn1998bw : @xcite ; sn2002ap : @xcite ) . the typical delay between the sn explosion and time of our observations is about 20 days , with four sne observed with a delay of over @xmath32 days .
in addition , three of the sne are embedded in host galaxies with strong radio emission . for these cases ,
we adopt upper limits that correspond to the brightness of the galaxy ( at least ten times the root - mean - square noise of the individual image ) .
four of the thirty three sne have been detected .
thus the detection rate of our experiment with a typical flux density limit of 0.15mjy ( @xmath33 ) is about @xmath34 .
figure [ fig : lcs ] provides a succinct summary of the radio light curves of the type ib / c sne .
two strong conclusions can be immediately drawn from this figure .
first , sne as bright as sn1998bw are rare .
second , there is significant dispersion in the luminosities of type ib / c sne , ranging from @xmath35 erg s@xmath8 hz@xmath8 at the bright end ( sn1998bw ) to that of sn2002ap which is fainter by about four orders of magnitude @xcite .
it is curious that sn2002ap also happens to be the nearest ib / c sn in our sample ( table [ tab : data ] ) .
six of the eight type ib / c sne detected in the radio to date cluster in the range of about @xmath36 erg s@xmath8 hz@xmath8 .
this may be partly due to a selection effect since the typical detection threshold is about @xmath37 erg s@xmath8 hz@xmath8 .
we also find that 28 of the 29 non - detections are no brighter than 0.1 times the luminosity of sn1998bw .
sn2002cg appears potentially brighter than sn1998bw only because it is embedded in a radio bright host galaxy ; we are therefore forced to use a @xmath38 limit on its luminosity ( table [ tab : data ] ) .
thus , the incidence of bright events like sn1998bw is @xmath1 . as with the radio luminosities , the peak times also exhibit great variation : at 1.4 ghz the emission from sn2002ap peaked at about 7 days , while for sn2002cj it peaked at about 65 days .
for sn1998bw the initial peak occurred at 15 days , followed by a second peak at about 40 days . similarly , at 8.5 ghz , sn1998bw peaked at 12 and 30 days past explosion , sn1983n peaked at about 30 days , and sn2002ap is predicted to have peaked at @xmath39 days ( the first observation at this frequency was taken about 4 days after the sn explosion ) .
if the radio emission arises from a synchrotron spectrum peaking at the self - absorption frequency , @xmath40 , then the peak time and peak luminosity directly measure the mean expansion speed @xcite .
this is simply because the self - absorption frequency is sensitive to the size of the source , while the luminosity is sensitive to the swept - up mass .
we use equation 16 of @xcite to evaluate the average expansion velocities : @xmath41 here , @xmath42 erg s@xmath8 hz@xmath8 is the peak luminosity , @xmath43 days is the time of peak emission relative to the sn explosion , and @xmath44 ghz is the peak frequency . we infer velocities ranging from @xmath45 to @xmath46 km s@xmath8 ( figure [ fig : vel ] ) . again , as with the luminosities , sn1998bw with @xmath47 is an exception .
we note that if free - free absorption plays a significant role , then @xmath48 is only an upper limit to @xmath40 , and @xmath49 is a lower limit to the intrinsic peak luminosity . in this case
, the inferred values of @xmath50 listed above will in fact be a lower limit to the actual expansion velocity .
however , this is probably not significant for type ib / c sne since their compact progenitors have high escape velocities and therefore fast winds and low circumburst densities . indeed , there is no evidence for free - free absorption either for sn1998bw @xcite or sn 2002ap @xcite . estimating the expansion velocities for the non - detections is not straightforward since we can
not ensure that the limits constrain the peak luminosity .
we are therefore forced to make an additional assumption .
for example , if we assume that most type ib / c sne are similar in their emission properties to sn1983n , then the majority of upper limits approximately sample the peak emission and the inferred upper limits are @xmath51 ( figure [ fig : vel ] ) . on the other hand ,
if the typical peak time is only a few days then our observations only constrain the decaying portion of the lightcurve and the expansion velocity may be higher . fortunately , this is not a significant problem based on the following argument .
the equipartition energy directly depends on the peak luminosity , @xmath52 erg ( see [ sec:2002cj ] ) , where @xmath53 erg s@xmath8 is the peak luminosity at 8.5 ghz . with a typical fading rate of @xmath54 in the optically thin regime , there are a few sne that could have reached a peak luminosity of the order of @xmath55 erg s@xmath8 if @xmath56 day post explosion .
this is a reasonable minimum peak time taking into account the deceleration time of the ejecta .
thus , the equipartition energy is at most @xmath57 erg , about two orders of magnitude lower than typical grbs ( [ sec : conc ] ) . for most non - detections
the limit is in fact much lower , @xmath58 to @xmath59 erg .
this indicates that a few of the non - detected sources may have in fact produced mildly relativistic ejecta , but these would still be energetically uninteresting when compared to sn1998bw let alone grb afterglows . in the previous section we found that no sn observed to date is comparable to sn1998bw especially in regard to the mean expansion speed .
sn1998bw was also interesting because it possessed an unusually large amount of energy in mildly relativistic ejecta .
however , a purely hydrodynamic explosions can also produce _ some _ amount of relativistic ejecta .
the energy of such ejecta can be estimated using well understood models of shock propagation in the pre - supernova cores @xcite .
the key parameters are @xmath2 and @xmath3 which can be inferred from the optical lightcurves and spectra using hydrodynamic models of a sn explosion in a co core coupled with radiative transfer calculations ( e.g. @xcite ) . in this section
, we investigate whether any of the detected type ib / c sne possess such large energy in high velocity ejecta that can not be explained by the simplest hypothesis of a purely hydrodynamic explosion . to this end , in table [ tab : ejecta ] we summarize the results of hydrodynamic models for the sne that have been detected in the radio . the ejecta produced in a hydrodynamic explosion has a density profile that can be described by power laws at low and high velocities , separated by a break velocity , which for type ib / c progenitors is given by @xcite : @xmath60 here @xmath61 erg and @xmath62 m@xmath11 . for typical values of @xmath2 and @xmath3 , the radio emission from the detected sne
is produced by ejecta above the break velocity .
in particular , for sn2002ap , @xmath63 km s@xmath8 , which is lower than the velocity of the ejecta producing the radio emission , @xmath64 km s@xmath8 @xcite . similarly , for sn1998bw @xmath65 ranges from about @xmath66 to @xmath67 km s@xmath8 ( depending on which model is assumed , table [ tab : ejecta ] ) while the radio emission was produced by ejecta expanding with @xmath68 .
the ejecta velocity profile extends up to a cutoff determined by significant radiative losses when the shock front breaks out of the star . for a radiative stellar envelope
this is @xmath69 @xcite , assuming a stellar radius of 1 r@xmath11 . for the sne considered here we find cutoff velocities of @xmath70 . to determine whether there is sufficient energy in fast ejecta to account for the radio observations we calculate the energy above a velocity , @xmath71 @xcite : @xmath72 where @xmath73 is the velocity in units of @xmath46 km s@xmath8 .
unfortunately , as can be seen from table [ tab : ejecta ] , only four ( including sn1998bw ) sne have sufficient optical data which is necessary to estimate @xmath2 and @xmath3 . of this limited sample ,
much of the radio data for sn1994i remain unpublished .
thus we are left with sn2002ap , sn1983n , and sn1998bw .
using the parameters given in table [ tab : ejecta ] for sn2002ap , @xcite find @xmath74 erg .
in contrast , from the radio observations we estimate @xmath75 erg . thus , there is no need , nor indeed room , for mildly relativistic ejecta in this sn .
we therefore disagree with the claims of high velocity jets carrying a large amount of energy , @xmath76 foe , made by @xcite and @xcite .
furthermore , the large discrepancy between the amount of energy inferred from the hydrodynamic models and the radio observations suggests that either the optically - derived parameters are in error , equation [ eqn : energy ] has an incorrect pre - factor , or the radio estimate is incorrect .
however , the radio estimate is relatively robust ( eventually related to equipartition energy estimates ) . on the other hand , as with sn1998be @xcite the total kinetic energy may have been over - estimated , possibly as a result of neglecting a mild asymmetry .
for sn1984l we do not have a direct estimate of the energy in the radio - emitting ejecta since the peak of the radio emission has been missed .
however , based on the similarity to sn1983n in the optically thin regime we estimate @xmath77 erg s@xmath8 hz@xmath8 .
this translates to a peak flux of @xmath78 mjy at the distance of sn1984l ( @xmath79 mpc ) . using the equipartition analysis presented in [ sec:2002cj ]
we estimate an energy of about @xmath80 erg , and an average expansion velocity of about @xmath81 . from equation [ eqn : energy ]
we find @xmath82 erg similar to the conundrum discussed above for sn2002ap . for sn1998bw ,
on the other hand , we find @xmath83 erg using the parameters inferred by @xcite , or @xmath84 erg using the parameters given by @xcite . in both cases ,
the energy available in fast ejecta is significantly lower than the energy inferred from the radio emission , @xmath85 erg . to conclude , for sn1984l and sn2002ap a hydrodynamic explosion can supply the energy and velocity that are responsible for the observed radio emission .
most likely , the same is true for the non - detections .
on the other hand , sn1998bw is a clear exception , exhibiting a significant excess of energy in ejecta moving with @xmath86 compared to what is available from a hydrodynamic explosion .
in the previous section we investigated the radio properties of type ib / c sne and found that in every respect sn1998bw was unique . in this section we compare the ib / c sample ( including sn1998bw ) with the radio afterglows of grbs . in figure [ fig : lcs2 ] we plot the radio lightcurves of grb970508 @xcite , a typical cosmological burst , and the nearest event , grb 030329 @xcite , in addition to the sn lightcurves .
as demonstrated by this figure and figure [ fig : hist1 ] , the radio lightcurves of grb afterglows and sne are dramatically different . furthermore , sn 1998bw is unique in both samples : it is fainter than typical radio afterglows of grbs but much brighter than type ib / c sne
. figures [ fig : lcs2 ] and [ fig : hist1 ] have significant implications , namely _ none _ of the type ib / c sne presented here could have given rise to a typical @xmath0-ray burst .
it has been suggested that grbs are distant type ib / c sne but with their jets pointed at the observer , whereas such a bias is absent in the nearby type ib / c sample .
however , most of our radio observations are obtained on a timescale of 10100 days ( see figure [ fig : lcs ] ) . scaling from the observed `` jet '' break times of a few days in grb afterglows , off - axis collimated explosions become spherical on a timescale of @xmath87 days @xcite at which point the relative geometry between
the observer and the explosion is not important .
thus , we find no evidence suggesting that all or even a reasonable majority of type ib / c sne give rise to grbs .
we now quantify the difference between the ib / c and grb samples .
our goal here is to determine the differential luminosity distribution , @xmath88 , which agrees with both detections and upper limits ; here @xmath88 is the number of events with luminosity between @xmath89 and @xmath90 .
it is important to include non - detections since they represent the majority of the sn data .
similarly , we include upper limits on the radio luminosity of grb afterglows that have been localized in other wave - bands ( i.e. optical and x - rays ) and for which a redshift has been measured .
unfortunately , as many as half of the grbs localized in the x - rays do not have a precise position , and hence a redshift .
for these afterglows it is not possible to provide a limit on the radio luminosity .
still , with a typical flux limit of about 0.3 mjy ( @xmath91 ; @xcite ) , and assuming that these sources have a similar distribution of redshifts to the detected afterglows , we find typical luminosity limits of about @xmath92 erg s@xmath8 hz@xmath8 , consistent with the peak of the distribution of detected afterglows . therefore , unless these sources are biased to low redshift we do not expect a strong bias as a result of neglecting them .
the quality of fit for @xmath88 is determined using the likelihood function , @xmath93 , with @xcite : @xmath94 where @xmath95 is the total number of sources ( sne or afterglows ) , and @xmath96 is a normalized gaussian profile centered on the observed luminosity of a detected source and with a width equal to the @xmath97 rms uncertainty in the luminosity .
we consider four models for @xmath88 based on the apparent distribution of the detections and upper limits : a gaussian , @xmath98 } , \label{eqn : gauss}\ ] ] a decreasing power - law , @xmath99 an increasing power - law , @xmath100 and a flat distribution , @xmath101 in each case we fit for the two free parameters ( e.g. @xmath102 and @xmath103 in equation [ eqn : gauss ] ) .
we do not use the increasing power law model for the individual distributions since the observations are clearly inconsistent with such a model .
the resulting best - fit models are shown in figure [ fig : hist2 ] and summarized in table [ tab : bayes ] .
we find that the sn population is modeled equally well with the gaussian , flat , or decreasing power law distributions , while the grb afterglows can be fit with a gaussian or flat distributions ; a decreasing power law provides a much poorer fit .
regardless of the exact distribution the two populations require distinctly different parameters , with a minimal overlap at the tails of the distributions .
fitting the combined sn and afterglow data with the models provided above ( figure [ fig : hist2 ] and table [ tab : bayes ] ) we find that even the best models ( an increasing power - law or a flat distribution ) provide a much poorer fit ; the likelihood of the fits are @xmath104 compared to the combined value of @xmath105 for the separate gaussian fits .
thus , the two populations can not be accommodated with a simple single distribution .
this points to a separate origin for the grb and type ib / c sn populations .
however , sn1998bw can be accommodated in either population .
it is equally plausibe that it is a low luminosity grb or the brightest type ib / c radio supernova known to date .
we presented vla radio observations of 33 type ib / c sne observed between late 1999 and the end of 2002 .
four of these sne have been detected , giving a detection rate of about @xmath34 above a typical @xmath33 flux limit of 0.15 mjy . at the same time , the combined detections and non - detections indicate that at most @xmath106 of type ib / c sne are as luminous as sn1998bw , although the single source which may be brighter is only so because it is embedded in a radio bright host galaxy .
we infer typical velocities of the radio - emitting ejecta of about @xmath107 km s@xmath8 for the detected sne , and upper limits in the same range for the non - detections .
we also find that a hydrodynamic explosion can supply the energy carried by the fastest ejecta . finally ,
none of the detected sne show clear evidence for variable energy input ( shells with different velocity or continued activity by the central engine ) ; however , we note that our sampling is quite sparse .
the measurements ( radio lightcurves ) and inferences ( energy in fast ejecta , energy addition ) offer no compelling reason to conclude that any of our sne have the special properties of sn1998bw ( [ sec : bw ] ) .
@xcite has proposed , based on the empirical lag - luminosity relation , that 25% of type ib / c sne are similar to sn1998bw .
we also compared the type ib / c sample with the sample of radio afterglow of grbs .
empirically , these two populations appear to be quite disparate .
this conclusion is reinforced by careful modeling of the luminosity distributions .
still , sn1998bw may belong to either population .
our four year survey of type ib / c sne was first and foremost motivated by the peculiar object , sn1998bw .
this supernova showed three attributes unique to grbs : relativistic ejecta , substantial reservoir of energy in such ejecta , and energy addition . a singular hydrodynamic explosion can not account for these attributes .
a natural explanation is that sn1998bw , like grbs , was , driven by an engine powerful enough to significantly modify the explosion .
our survey has demonstrated that sn1998bw - like events are rare in the local sample .
this begs the question : what is sn1998bw ?
two popular scenarios have been suggested .
the first the `` off - axis '' scenario holds that sn1998bw is a typical grb albeit nearby and with collimated ejecta pointed away from us @xcite .
this hypothesis is attractive because of its simplicity .
we know grbs exist and most of them do not point towards us @xcite
. in the other scenario sn1998bw is a new type of explosion with little energy in ultra - relativistic ejecta @xcite .
evidence in favor of this idea is best illustrated by figure [ fig : hist3 ] where we find that grb980425 is consistently at the faint end of the grb population .
unfortunately , we are not able to decisively resolve this controversy .
as demonstrated by figure [ fig : hist1 ] , one could argue that sn1998bw is at the bright end of the radio luminosity function of type ib / c supernovae or at the faint end of grb radio afterglow .
the expected rate of sn1998bw - like events in the off - axis framework is about @xmath108 of type ib / c events , given the average beaming factor of about @xmath109 derived by @xcite .
thus , it is not entirely improbable that one out of about 40 type ib / c sne observed to date is an off - axis grb . as an aside
, we can use our @xmath106 limit and compare the event rate of type ib / c sne with the observed rate of grbs ( [ sec : intro ] ) to place a limit of @xmath110 on the beaming fraction .
this corresponds to a limit of @xmath111 on the jet opening angles of grbs ; narrower jets are not likely .
this result may also be interpreted as a limit on angular size of the highly relativistic core in models of variable energy and/or lorentz factor across the surface of the jet @xcite . in the gsn framework , we now know that at most a few percent of type ib / c are possibly gsne . at the same time , the recent grb030329 was accompanied by a sn similar to sn1998bw ( sn2003dh ; @xcite ) .
thus , an investigation of the number and properties of gsne requires observations of both local type ib / c sne and grbs . while we can not determine the exact origin of sn1998bw based on the statistics of our survey ,
the ultimate detection of similar events at the level of about @xmath14 may in fact allow us to distinguish between the off - axis and gsn scenarios .
the discovery of broad optical lines in sn1998bw and large explosive energy release , @xmath112 foe , prompted some astronomers to use the designation `` hypernovae '' for sn1998bw - like sne .
unfortunately , this designation is not well defined . to begin with
, the term hypernova was first used by @xcite to describe the grb / afterglow phenomenon ; thus , this term implies a connection to grbs .
the prevalent view now is that hypernovae are characterized by broad optical absorption lines and larger than normal energy release .
however , neither of these criteria has been defined quantitatively by their proponents . ignoring this important issue ,
the following have been suggested to be hypernovae : the type ib / c sne 1992ar @xcite , 1997dq @xcite , 1997ef @xcite , 1998ey @xcite , and 2002ap @xcite , and the type ii sne 1992am @xcite , 1997cy ( e.g. @xcite ) , and 1999e @xcite . some have also been claimed to be associated with grbs detected by batse , but at a low significance .
our view is that the critical distinction between an ordinary supernova and a grb explosion are relativistic ejecta carrying a considerable amount of energy .
such ejecta are simply not traced by optical spectroscopy .
this reasoning is best supported by the fact that the energy carried by the fast ejecta in sn1998bw and sn2002ap @xcite differ by four orders of magnitude even though both exhibit broad spectral features at early times .
thus , broad lines do not appear to be a good surrogate for sn1998bw - like objects .
in addition , in two cases , sne 2002ap and 1984l , the energy inferred from the radio observations indicates that the total kinetic energy as inferred by optical spectroscopy and light curves may have been over - estimated by an order of magnitude ( [ sec : energy ] ) .
it is possible that `` hypernovae '' are in fact only slightly more energetic than typical type ib / c sne , but exhibit a mild degree of asymmetry , leading to excessively high estimates of the total energy when a spherical explosion is assumed .
we suggest that the term hypernova be reserved for those sne , like sn1998bw , which show direct evidence for an engine through the presence of relativistic ejecta . as illustrated by sn1998bw , the relativistic ejecta
are reliably traced by radio observations .
we end with the following conclusions .
first , radio observations provide a robust way of measuring the quantity of energy associated with high velocity ejecta .
this allows us to clearly discriminate between engine - driven sne such as sn1998bw and ordinary sne , powered by a hydrodynamic explosion , such as sn2002ap @xcite and sne 2001b , 2001ci , and 2002cj presented here .
second , since at least @xmath113 of local type ib / c sne are not powered by inner engines and furthermore have a total explosive yield of only @xmath59 erg in fast ejecta ( where fast means @xmath114 compared to @xmath115 in grbs ) , there is a clear dichotomy between type ib / c sne and cosmic , engine - driven explosions ( figure [ fig : hist3 ] ) .
the existence of intermediate classes of explosions and the nature of sn1998bw may be ascertained with continued monitoring of several hundred type ib / c sne and cosmological explosions .
fortunately , such samples will likely become available over the next few years .
we thank b. schmidt for helpful comments .
grb and sn research at caltech is supported in part by funds from nsf and nasa .
sn2001b was discovered in images taken on 2001 , jan 3.61 and 4.57 ut , approximately 5.6 arcsec west and 8.9 arcsec south of the nucleus of ic391 @xcite .
the sn explosion occurred between 2000 , dec 24.54 ut and the epoch of discovery .
based on an initial spectrum , taken on 2001 , jan 14.18 ut @xcite concluded that the sn was of type ia approximately @xmath116 days past maximum brightness , showing well defined siii and caii features , with a siii expansion velocity of about 7400 km sec@xmath8 . a subsequent spectrum taken on 2001
, jan 23 indicated that the sn was in fact of type ib based on clear hei absorption lines @xcite .
we observed sn2001b on 2001 , feb 4.49 ut at 8.46 ghz .
an initial analysis revealed a source which was interpreted as the host galaxy of the sn , with a flux of @xmath117 mjy .
a second epoch obtained on 2002 , oct 28.45 ut revealed that the source has faded below @xmath118 mjy , establishing it as the radio counterpart of sn2001b .
( figure [ fig : lcs ] ) .
sn2001ci was initially detected in images taken with the katzman automatic imaging telescope on 2001 apr 25.2 ut , 6.3 arcsec west and 25.4 arcsec north of ngc3079 @xcite .
these observations did not provide conclusive evidence that the source was in fact a sn . a spectrum obtained by @xcite on 2001 , may 30 ut revealed that the source was in fact a type ic sn about @xmath119 weeks past maximum brightness @xcite .
we observed the sn on 2001 , jun 6.35 ut at 8.46 ghz , but did not detect the source since it appeared to be part of the host galaxy extended structure . a second epoch taken on 2002 ,
jun 10.9 ut revealed a clear fading at the optical position of the sn with a flux of @xmath120 mjy in the first epoch , and a @xmath121 limit of @xmath122 mjy in the second epoch .
sn2002cj was discovered on 2002 , apr 21.5 ut , 1.4 arcsec west and 3.9 arcsec south of the nucleus of eso 582-g5 , at a distance of 106 mpc @xcite .
the sn explosion occurred between apr 9.5 ut and the epoch of discovery .
spectra of the sn obtained on 2002 , may 2.43 and may 7 ut revealed that sn2002cj is of type ic @xcite .
we initially observed the sn on 2002 , jun 3.19 ut at 1.43 , 4.86 , and 8.46 ghz and detected a faint source at all three frequencies .
the sn spectrum is given by @xmath123 between 1.43 and 4.86 ghz , and @xmath124 between 4.86 and 8.46 ghz , indicating that the spectral peak was most likely located between 1.43 and 4.86 ghz during the first epoch ( @xmath125 days ) .
in subsequent observations at 1.43 ghz the sn brightened and then faded , as expected if the peak was in fact above 1.43 ghz initially , and shifted through the band at later epochs .
using the expected shape of the spectrum , @xmath126 for @xmath127 and @xmath128 for @xmath129 ( with @xmath130 ) , we find @xmath131 mjy at @xmath132 ghz and @xmath133 days .
we use these values along with the well - established equipartition analysis @xcite to derive some general constraints on the properties of the emitting material .
in particular , the energy of a synchrotron source with flux density , @xmath134 , can be expressed in terms of the equipartition energy density , @xmath135 where @xmath136 , @xmath137 @xmath138as , @xmath139 erg , and @xmath140 and @xmath141 are the fractions of energy in the electrons and magnetic fields , respectively . in equipartition @xmath142 and
the energy is minimized ; a deviation from equipartition will increase the energy significantly . for sn2002cj
we find @xmath143 @xmath138as ( i.e. @xmath144 cm ) , which indicates an average expansion velocity , @xmath145 .
the equipartition energy is @xmath146 erg , indicating a magnetic field strength of @xmath147 g. llllllcccc 1999ex & 7310 & oct 25.6nov 9.5 & nov 18.02 & 8.523.4 & 54 & no & @xmath148 & @xmath149 & + + 2000c & 7348 & 1999 dec 302000 jan 8.3 & feb 4.04 & 26.736.0 & 59 & hint & & @xmath150 & + 2000f & 7353 & 1999 dec 30.3jan 10.2 & & & 92 & + 2000s & 7384 & @xmath151feb 28@xmath152 & & & 94 & + 2000cr & 7443 & jun 21.225.9 & jul 3.04 & 7.111.8 & 54 & no & @xmath153 & & + 2000de & 7478 & jun 6aug 10.9 & & & 39 & + 2000ds & 7507 & may 28oct 10.4 & & & 21 & + 2000dt & 7508 & sep 21oct 13.1 & & & 107 & + 2000dv & 7510 & @xmath151oct 17.1 & & & 63 & + 2000ew & 7530 & may 8nov 28.5 & 2002 jun 26 & 575780 & 15 & no & & & @xmath154 + 2000fn & 7546 & nov 119 & dec 29.22 & 40.258.2 & 72 & no & @xmath155 & & + + 2001b & 7555 & 2000 dec 24.5jan 3.6 & feb 4.49 & 31.942.0 & 24 & yes & @xmath156 & & + & & & 2002 oct 28.45 & 672.9683 & & no & @xmath157 & & + 2001 m & 7568 & jan 3.421.4 & feb 4.45 & 14.132.1 & 56 & no & @xmath158 & & + 2001ai & 7605 & mar 19.428.4 & mar 31.36 & 312 & 118 & no & @xmath159 & @xmath160 & @xmath160 + 2001bb & 7614 & apr 15.329.3 & may 5.26 & 620 & 72 & no & @xmath161 & @xmath162 & @xmath163 + 2001ch & 7637 & 2000 nov 28.3may 28.5 & & & 46 & + 2001ci & 7618 & apr 17.225.2 & jun 6.35 & 42.250.2 & 17 & yes & @xmath164 & & + & & & 2002 jun 10.90 & 411.7419.7 & & no & @xmath165 & & + 2001ef & 7710 & aug 29sep 9.1 & sep 14.3 & 5.216.3 & 38 & no & @xmath166 & & + & & & 2002 jun 18.0 & 281.9293 & & no & & & @xmath167 + 2001ej & 7719 & sep 117.1 & sep 27.25 & 10.626.7 & 63 & no & @xmath168 & & + & & & 2002 jun 14.0 + 18.0 & 269.9469 & & no & & & @xmath169 + 2001em & 7722 & sep 10.315.3 & & & 91 & + 2001eq ( ? ) & 7728 & aug 31.3sep 12.3 & & & 118 & + 2001fw & 7751 & oct 26nov 11.2 & & & 139 & + 2001fx & 7751 & oct 14.2nov 8.2 & & & 126 & + 2001is & 7782 & dec 1423 & 2002 jun 14.0 + 18.0 & 175184 & 60 & no & & & @xmath170 + + 2002j & 7800 & jan 9.421.4 & jun 18.0 & 147.6159.6 & 58 & no & & & @xmath169 + 2002ap@xmath171 & 7810 & jan 28.5 & feb 1.03 & 3.5 & 7 & yes & @xmath172 & & + 2002bl & 7845 & feb 14mar 2.9 & mar 8.26 & 622 & 74 & no & @xmath161 & @xmath173 & + 2002bm & 7845 & jan 16mar 6.2 & jun 26.0 & 111.8161 & 85 & no & & & @xmath174 + 2002cg & 7877 & mar 28.5apr 13.5 & jun 26.0 & 73.589.5 & 150 & no & & & @xmath175 + 2002cj & 7882 & apr 9.521.5 & jun 3.19 & 42.754.7 & 106 & yes & @xmath176 & @xmath177 & @xmath178 + & & & jun 18.0 & 57.569.5 & & yes & & & @xmath179 + & & & jun 26.0 & 65.577.5 & & yes & & & @xmath180 + 2002cp & 7887 & apr 11.228.2 & jun 1.96 & 34.851.8 & 80 & no & @xmath181 & @xmath182 & @xmath183 + 2002cw & 7902 & 2001 oct 1.2may 16.5 & & & 71 & + 2002dg & 7915 & 2002 may 5may 31.3 & jul 9.95 & 39.766 & 215@xmath184 & hint & @xmath185 & @xmath186 & @xmath187 + 2002dn ( ? ) & 7922 & may 31.5jun 15.5 & jul 4.4 & 18.933.9 & 115 & no & @xmath188 & @xmath189 & @xmath190 + 2002dz & 7935 & 2001 nov 10.2jul 16.5 & & & 84 & + 2002ex & 7964 & aug 19.3 & & & 180 & + 2002fh ( ? ) & 7971 & apr 9may 9.3 & & & 1870 & + 2002ge(?)@xmath191 & 7987 & 2000 oct 1.82002 oct 7.9 & oct 12.0 & @xmath192 & 47 & no & @xmath193 & @xmath194 & + 2002gy & 7996 & oct 9.416.4 & oct 28.45 & 19.126.1 & 114 & no & @xmath183 & @xmath195 & + 2002hf & 8004 & oct 22.329.3 & nov 7.17 & 8.916.9 & 88 & no & @xmath157 & & + 2002hn & 8009 & oct 21.530.5 & nov 7.42 & 7.916.9 & 82 & no & @xmath196 & & + 2002ho & 8011 & may 27nov 5.1 & nov 15.33 & @xmath197 & 42 & no & @xmath198 & @xmath199 & + 2002hy & 8016 & oct 13.1nov 12.1 & nov 21.71 & 9.639.6 & 58 & no & @xmath200 & & + 2002hz & 8017 & nov 2.212.2 & 2003 jan 21.09 & 69.970.9 & 85 & no & @xmath201 & & + 2002ji & 8025 & apr 10nov 30.8 & dec 5.67 & 38.459.4@xmath202 & 23 & no & @xmath189 & & + 2002jj & 8026 & oct 1.424.4 & dec 15.28 & 51.974.9 & 66 & no & @xmath185 & & + 2002jp & 8031 & may 14.2nov 23.5 & dec 14.55 & @xmath203 & 58 & no & @xmath204 & & + 2002jz & 8037 & 2001 dec 52002 dec 23.3 & 2003 jan 3.28 & @xmath205 & 24 & no & @xmath206 & & + lcccccc 1984l & @xmath207 & @xmath208 & @xmath209 & & & 1 + & & @xmath210 & & & & 2 + & & & @xmath211 & & @xmath212 & 3 + 1994i & @xmath213 & @xmath214 & 0.07 & @xmath215 & @xmath216 & 4 + & & @xmath217 & 0.07 & @xmath218 & & 5 + & 1 & & & & & 6 + 1998bw & 30 & & 0.7 & 13.8 & 40 & 7 + & 50 & 10 & 0.4 & 13.8 & 40 & 8 + & 22 & & 0.5 & 6.5 & & 9 + & 2 & 2 & 0.2 & & & 10 + & & & @xmath219 & & & 11 + 2002ap & @xmath220 & @xmath221 & 0.07 & 5 & @xmath222 & 12 llcc sn & gaussian & ( 26.1 , 1.0 ) & 22.5/36 + sn & d. powerlaw & ( @xmath223 , 25.4 ) & 22.4/36 + sn & flat & ( 20.0 , 29.1 ) & 22.3/36 + grb & gaussian & ( 31.0 , 0.8 ) & 38.8/33 + grb & d. powerlaw & ( @xmath224 , 29.6 ) & 48.9/33 + grb & flat & ( 29.6 , 32.4 ) & 37.7/33 + sn+grb & gaussian & ( 28.4 , 2.4 ) & 130.3/71 + sn+grb & d. powerlaw & ( @xmath225 , 25.3 ) & 123.8/71 + sn+grb & i. powerlaw & ( 3.7 , 32.3 ) & 103.7/71 + sn+grb & flat & ( 22.6 , 32.4 ) & 104.9/71 + | the association of @xmath0-ray bursts ( grbs ) and core - collapse supernovae ( sne ) of type ib and ic was motivated by the detection of sn1998bw in the error box of grb980425 and the now - secure identification of a sn1998bw - like event in the cosmological grb030329 .
the bright radio emission from sn1998bw indicated that it possessed some of the unique attributes expected of grbs , namely a large reservoir of energy in ( mildly ) relativistic ejecta and variable energy input .
the two popular scenarios for the origin of sn1998bw are a typical cosmological burst observed off - axis or a member of a new distinct class of supernova explosions ( gsne ) . in the former , about 0.5% of local type ib / c
sne are expected to be similar to sn1998bw ; for the latter no such constraint exists . motivated thus , we began a systematic program of radio observations of most reported type ib / c sne accessible to the very large array .
of the 33 sne observed from late 1999 to the end of 2002 at most one is as bright as sn1998bw . from this
we conclude that the incidence of such events is @xmath1 .
furthermore , analysis of the radio emission indicates that none of the observed sne exhibit clear engine signatures .
finally , a comparison of the sn radio emission to that of grb afterglows indicates that none of the sne could have resulted from a typical grb , independent of the initial jet orientation .
thus , while the nature of sn1998bw remains an open question , there appears to be a clear dichotomy between the majority of hydrodynamic and engine - driven explosions . |
phenomena in high dimensions have been the subject of many studies . in particular
, one may be interested in studying higher dimensional geometric objects called polytopes .
a @xmath3 @xmath4 is a bounded subset of @xmath1 formed by taking the intersection of finitely many halfspaces .
the @xmath2-vector of a polytope @xmath5 is a vector @xmath6 such that @xmath7 is the number of @xmath8-dimensional faces of @xmath5 . in particular
, @xmath9 is the number of vertices of @xmath5 and @xmath10 is the number of @xmath11-dimensional faces , known as @xmath12 .
there are some known results on the distribution of the @xmath2-vector of types of random polytopes .
previous work has found that individual components of the @xmath2-vectors are approximately gaussian when the polytopes are generated according to various distributions . in particular , brny and vu
@xcite proved that if a polytope is taken as the convex hull of independent and identically distributed ( i.i.d . )
points according to the standard gaussian distribution , then the volume and individual components of the @xmath2-vector of that polytope satisfy a central limit theorem ; that is , converge to a gaussian distribution as the number of i.i.d .
random points tends to infinity .
brny and reitzner @xcite proved central limit theorems for the volume and components of the @xmath2-vector of a so - called poisson random polytope ; that is , the convex hull of the intersection of an arbitrary volume 1 convex body @xmath13 with a poisson process @xmath14 . following earlier work of reitzner @xcite ,
vu @xcite proved that the volume and components of the @xmath2-vector of a random polytope drawn from points inside a smooth convex set also satisfy central limit theorems . in this paper , we consider the joint distributions of the @xmath2-vectors of random polytopes .
the method used in this paper for constructing a random polytope is taking the convex hull of a set of i.i.d .
random points generated from a fixed distribution .
our conjecture is the following : _ let n be a positive integer .
let @xmath15 be a collection of random points @xmath16 , which are independent and identically distributed according to a fixed distribution .
letting @xmath17 be the convex hull of @xmath18 , under mild assumptions on the distribution of the @xmath19 , the joint distribution of the @xmath2-vector satisifes a central limit theorem_. in the five underlying distributions of the random points considered , numerical evidence supports the conjecture .
the metric used to compare the closeness of the distribution of the @xmath2-vectors with the gaussian distribution is based on the kolmogorov distance , @xmath20 . for now one may think of @xmath20 as a tool that measures the distance between two distributions and that @xmath21 .
if @xmath20 is a small value then the two distributions are considered to be close to each other .
the values of @xmath20 presented in the table are for comparison to the appropriate gaussian distribution ; details are discussed in section 3 . here
, @xmath22 is the dimension , @xmath0 is the number of random points , and @xmath23 is the sample size or number of @xmath2-vectors .
+ .uniform distribution in cube [ cols="^,^,^,>,<",options="header " , ] [ table : gaussian ] since @xmath20 is a metric that takes on values in the interval [ 0,1 ] , a value with order of magnitude @xmath24 as the kolmogorov distance should be considered a rather small value .
figure 1 contains histograms of individual components of 64,000 @xmath2-vectors computed from the convex hull of 25,000 i.i.d .
random points uniformly distributed in the 5-dimensional cube . after standardizing the @xmath2-vectors to have sample mean zero and identity sample covariance , the resulting data lie in @xmath25 . as further illustration ,
the 2-dimensional data of the standardized @xmath2-vector is plotted in @xmath26 .
-vectors in @xmath27 .
data came from 25,000 @xmath2-vectors computed from random polytopes generated by 64,000 i.i.d .
uniformly distributed points in the 5-dimensional cube . ]
it was mentioned in the introduction that one way to construct a random polytope is to take the convex hull of random points . one way to generate random points
is to let @xmath28 be a collection of points generated independently according to a fixed distribution .
five distributions used to generate random points were considered and the random polytope studied was the convex hull of the @xmath28 .
the underlying distributions considered are the following : 1 . uniform in the cube .
define @xmath29 where @xmath30 .
a random point @xmath31 is constructed by picking each component of @xmath32 independently according to the uniform distribution in the interval [ 0,1 ] .
the resulting random point @xmath32 is uniformly distributed in @xmath33 .
2 . uniform in the euclidean ball .
define @xmath34 in the @xmath35 norm as @xmath36 where @xmath37 . to generate a random point @xmath38 such that @xmath32 is uniformly distributed in @xmath34 ,
first generate a point @xmath39 that is distributed according to the standard @xmath40 dimensional gaussian distribution .
this is done by picking each component of @xmath41 independently according to the standard univariate normal distribution .
let @xmath42 , where @xmath43 .
then it is classical that @xmath32 is uniformly distributed in @xmath34 .
see [ 1 ] for a proof .
3 . uniform in the @xmath44 ball .
define @xmath45 in the @xmath44 norm as @xmath46 where @xmath47 .
let @xmath48 where the components of @xmath41 are generated independently according to the distribution with the density function @xmath49 .
let @xmath50 , where @xmath51 ; it is proven in @xcite that @xmath32 is uniformly distributed in @xmath45 .
4 . the standard normal distribution . let @xmath52 be a random vector and generate each components of @xmath32 independently according to the standard normal distribution
. then @xmath38 is distributed as a standard normal random vector .
5 . uniform in the hemisphere .
similar to ( 2 ) , but take the absolute value of the first component of the random point @xmath32 in ( 2 ) .
note that unlike the underlying convex bodies in ( 1)-(3 ) , this body is neither smooth nor a polytope .
the software matlab and qhull were used to carry out the following simulations and computations .
one realization of the @xmath2-vector is computed by first fixing the dimension @xmath22 , generating @xmath0 i.i.d .
points @xmath53 from one of the fixed probability distributions described above , taking the convex hull of these points to construct a random polytope @xmath5 , and computing the @xmath2-vector of the random polytope . assuming the random polytope is _ simplicial _ ,
that is , each of its facets has exactly @xmath22 vertices , the @xmath2-vector of the polytope can be directly computed from the facets .
let @xmath54 be the @xmath55 facets of @xmath5 , where @xmath56 is the @xmath57th vertex of the @xmath8th facet , for @xmath58 , @xmath59 . by the definition of @xmath2 , @xmath60 .
since @xmath61 is the number of @xmath62-dimensional faces of @xmath5 , @xmath61 is found by counting all intersections of size @xmath11 between pairs of facets @xmath63 and @xmath64 , for @xmath65 . since
each @xmath62-dimensional face is the intersection of exactly two facets , this accounts for all @xmath62-dimensional faces .
for example , 3-dimensional faces @xmath66 and @xmath67 , where each @xmath68 is a vertex for that face , have intersection @xmath69 which is of size 2 so it must be an edge , but @xmath66 and @xmath70 have intersection @xmath71 which is of size 1 so it is not an edge .
let @xmath72 be these distinct intersections of size @xmath62 ; clearly @xmath73 . by keeping track of all the intersections of size @xmath11
, we obtain a list of all @xmath62-dimensional faces of @xmath5 .
we can then repeat this process with those faces . continuing inductively in this manner ,
the remaining components of @xmath2 can be determined .
note that the convex hull of points in general can be non - simplicial ( a facet may have more than @xmath22 vertices ) , and so the above algorithm for constructing the @xmath2-vector from the facets is not valid for all polytopes .
however , the probability of a non - simplicial convex hull arising from the distributions of points considered in this paper is 0 .
a collection of @xmath23 realizations of @xmath2-vectors was obtained , and the data were then standardized and compared to the standard gaussian distribution .
the following demonstrates the computations of standardizing the @xmath2-vectors .
let @xmath74 be the @xmath75 @xmath2-vector ; then @xmath76 is the @xmath77 component of the @xmath75 @xmath2-vector .
define the sample mean @xmath78 of the _ f_-vectors as @xmath79 where @xmath80 . given the sample data @xmath81 ,
the sample covariance matrix @xmath82_{i , j=1}^d$ ] is a @xmath22-by-@xmath22 matrix with entries given by @xmath83 . to standardize to identity sample covariance ,
first diagonalize the sample covariance @xmath84 , where @xmath85 is a diagonal matrix of eigenvalues and @xmath86 is an orthogonal matrix where the columns of @xmath86 are the eigenvectors .
moreover , the eigenvalues of @xmath85 can be assumed to be in decreasing order , @xmath87 .
note that the components of the @xmath2-vectors have some dependence on each other . for polygons ,
the components of the @xmath2-vector obey @xmath88 .
for convex polyhedra in @xmath26 , the components of the @xmath2-vector satisfy @xmath89 , known as euler s relation . in higher dimensions ,
linear dependence between the components of the @xmath2-vector causes the covariance matrix of the @xmath2-vector to be singular .
these type of linear dependencies are known as dehn - sommerville equations for simplicial polytopes [ 6 ] .
a singular covariance matrix leads to zero eigenvalues in the diagonalization of the covariance matrix . in practice ,
the distinction between non - zero eigenvalues and approximately zero eigenvalues of the sample covariance matrix was easy to identify .
in such case , let the eigenvalues of the sample covariance matrix that are close to zero be eliminated to ensure that @xmath90 is well - defined .
so define a new matrix @xmath91 where @xmath92 is a @xmath93-by-@xmath93 matrix with the eigenvalues of matrix @xmath85 that are close to zero eliminated .
also define a new matrix @xmath94 where the eigenvector @xmath95 corresponds to the @xmath96 eigenvalue .
then let @xmath97 , where @xmath98 is the @xmath75 re - centered @xmath2-vector .
then @xmath99 is now a standardized data set with mean 0 and identity sample covariance .
after standardization , one can now compare the standardized @xmath2-vectors to the standard gaussian distribution , in terms of the kolmogorov distance .
the kolmogorov distance between two random variables @xmath100 is defined to be @xmath101-{\mathbb{p}}[y\le t]|$ ] .
observe that since the standardized data are conjectured to be approximately gaussian , any linear combination of their components should also be close to gaussian .
this motivates the following numerical test .
fix @xmath102 , and let @xmath103 be @xmath102 independent samples of uniformly chosen points on the sphere in @xmath104 . for each @xmath105 , compute the kolmogorov distance @xmath106 where @xmath107 is the empirical cumulative distribution function of the data projected onto the direction @xmath105 ; that is , @xmath108 and @xmath109 is the cumulative distribution function of the standard normal distribution . finally , define a measure @xmath20 of the distance from the data to gaussian by @xmath110 the values presented in the tables in section 1 are obtained using @xmath111 .
the authors would like to thank advisor elizabeth meckes for her guidance and mentoring on the subject along with her help in editing this paper .
this work was supported by nsf grants dms-0905776 and dms-0852898 . | choose @xmath0 random , independent points in @xmath1 according to a fixed distribution .
the convex hull of these points is a random polytope . in some cases ,
central limit theorems have been proven for the components of @xmath2-vectors of random polytopes constructed in this and similar ways . in this paper
, we provide numerical evidence that the components of the @xmath2-vectors of random polytopes generated according to five different distributions are approximately jointly gaussian for large @xmath0 . |
it is known , since long time , that ring - shaped organic molecules , such as benzene or naphthalene , show very large diamagnetic susceptibilities and a very pronounced diamagnetic anisotropy : the diamagnetic susceptibility has normal values in the plane of the aromatic molecules but abnormally large ones in the direction perpendicular to this plane .
a first qualitative explanation of this phenomenon as arising from the larmor precession of electrons in orbits including many nuclei can be traced back to p. ehrenfest in 1925 @xcite .
this idea was further developed eleven years later by l. pauling who proposed a quasi - classical description of the @xmath0 electronic currents around the rings using an electrical network as a model @xcite .
one year latter , in 1937 , f. london set up the `` modern '' theoretical description of this phenomenon by using a tight binding model which included the magnetic field in form of a complex phase of the usual transfer integral @xcite .
this is the so - called london ( or peierls ) substitution widely used nowadays .
the comparison between the results given by these early theoretical works and the experimental data available at that time is quite satisfactory @xcite .
however , the interest for this topic has nearly vanished most probably due to experimental limitations . for instance , in the theory of london , the electronic spectrum of aromatic molecules is a periodic function of the magnetic flux with a period equal to a flux quantum @xmath1 . but the value of the magnetic field needed to cover a full period is of order @xmath2 t for the materials existing until few years ago .
such huge values can not be reached experimentally , and are due to the extremely small perimeter , of order a few tens of angstrms , of the materials studied previously .
today , material science has made considerable progress and many new carbon materials with nanoscopic characteristic length exist .
an example is given by multi - walled carbon nanotubes with a diameter of approximately 10 nm @xcite .
for these new materials , the magnetic field needed to reach a flux quantum is about a few tens of tesla only and the periodicity mentioned before can be observed @xcite .
on the other hand , during the last years considerable progress has been made in precise measurements of the real part of the low - frequency dielectric function @xmath3 .
in particular these experiments have been extended to ultra - low temperatures .
ratios of @xmath4 up to @xmath5 were achieved in the mk temperature regime . a rather spectacular success associated with
that progress was the observation of a strong magnetic - field dependence of the polarisability of multicomponent glasses @xcite .
these developments suggest reconsideration of the magnetic field effects in organic nanostructures such as large aromatic molecules or carbon nanotubes for instance . in particular , the magnetic field should have a very pronounced effect on the polarisability of these organic systems which should be observable nowadays @xcite . these measurements
could then be used as a new spectroscopic tool to obtain informations on the electronic properties of organic materials .
the aim of this paper is to review various theoretical works developed in dresden during the last two years in that direction @xcite .
similar ideas were already pointed out recently for quite different materials and purposes .
indeed , in @xcite small particles ( rings , disks or spheres ) were considered in connection with weak localisation . in particular
, it was found that the polarisability should be greater in an applied magnetic field than in zero field because a field partially suppresses weak localisation .
this theoretical prediction was observed very recently in the ac polarisability of mesoscopic rings @xcite .
although the studies already done until now on the behaviours of the dielectric susceptibility of cylindrical organic materials under magnetic field @xcite should be improved in several ways by including , e.g. , explicitely coulomb interactions , we believe the phenomena pointed out here are robust in the sense that they will survive a more sophisticated theoretical treatments .
therefore , we think that the polarisability in a magnetic field could be a useful quantity to study experimentally in order to obtain insight into the electronic structure of carbon made materials . in our opinion , this justify the topic of this review and , we hope that it will help to motivate new experiments .
the series of works described here is related to the studies of persistent currents in metallic rings . before starting
, we would like to mention recent review articles on this topic .
interested readers could refer to ref .
@xcite for the case of diffusive electron motion and to ref .
@xcite for ballistic electron motion .
the review is organised as follows . in section 2
we describe the aharonov - bohm effect in cylindrical nanostructures by using london s substitution . in section 3 and 4
, we point out dramatic consequences of this effect on the dielectric function of nanoscopic organic cylinders such as carbon nanotubes ( section 3 ) , and a finite plane of graphite where the edge states existing in these materials define effective rings ( section 4 ) . finally , before concluding , we point out a few properties of strongly interacting electrons on rings and cylinders in an applied field .
in this paper , we are interested in the effects of an applied magnetic field on the orbital degrees of freedom only .
therefore we consider spinless fermions except in the last section .
the zeeman interaction was briefly discussed in @xcite . in the range of field strength
we are interested in , the diamagnetic response is the most important one .
let us first consider a ring of @xmath6 sites pierced by an axial magnetic field .
the hamiltonian , making use of the london substitution @xcite , is the following @xmath7 where @xmath8 , ( @xmath9 ) are spinless fermion creation ( annihilation ) operators and @xmath10 is the magnetic flux through the ring in units of the flux quantum @xmath11 ( @xmath1 ) .
it is proportional to the area enclosed by the ring @xmath12 where @xmath13 is the lattice constant . for a ring with a few hundred sites and assuming @xmath14 , the magnetic field needed to reach @xmath15 is not more than a few tens tesla - values reachable experimentally . introducing the wave function @xmath16 , where @xmath17 are the @xmath0 orbitals located at sites @xmath18 ,
one obtains the following eigenvalue equations with periodic boundary conditions @xmath19where @xmath20 is the eigenvalue . by performing a pseudo - gauge transformation introduced by bayers and yang @xcite , @xmath21 , eqs .
( [ eigenring ] ) can be written in the form @xmath22 the effect of the magnetic field appears in the new periodic boundary conditions .
an analysis of them shows that the spectrum of the system is a periodic function of @xmath10 with period 1 , as already mentioned in the introduction .
this is an important information since it implies that every thermodynamic quantity is also periodic function of @xmath10 with the same period .
moreover , the system is invariant with respect to the transformation @xmath23 .
therefore its spectrum - and every thermodynamic quantity - is symmetric with respect to @xmath24 . in principle , it is then necessary to study the system only for @xmath25 $ ] .
the eigenvalues for the ring are @xmath26with @xmath27 . using this expression one notices crossing between levels occurring at special values of @xmath10 , namely @xmath28 and @xmath29 ( for an example see fig .
1 ) . let us now consider cylinders made by rolled square lattices with an applied axial magnetic field ( fig .
the hamiltonian for spinless fermions is @xmath30where the magnetic flux is still given by expression ( [ flux ] ) .
the resulting eigenvalues are @xmath31 with @xmath32 and @xmath33 .
we have applied open boundary conditions at the ends of the cylinder .
the resulting one - electron wave functions are @xmath34 where @xmath35 is the vacuum
. again , using the expression ( [ spectrum ] ) , one notices many level crossings induced by the magnetic field ( fig .
3 ) . compared to the results obtained for rings , the important difference is that the crossings appear for values of @xmath10 depending on the geometry of the system @xmath36 ; in particular , they appear for other values than @xmath37 , @xmath38 and @xmath29 . due to their intricate patterns , numerous changes of the ground state induced by the magnetic field
do occur . as an example
, we show in figure 4 as function of the magnetic flux the energies of the ground state and the first excited state of a cylinder with @xmath39 , @xmath40 and with @xmath41 electrons in it . the number of crossing points and the values of @xmath10 for which they occur depend crucially on the geometry of the system as we have already mentioned above but , also , on the electron filling .
we will not attempt to explain the crossing point pattern here . instead , as shown in the next section , we want to point out that the changes of the ground state associated with the crossing points , have strong influences on the static polarisability .
therefore this quantity could be used as a new spectroscopic tool to obtain informations on excited states or on the system geometry for instance .
carbon nanotubes form a class of cylindrical systems of growing importance nowadays @xcite .
the calculations above can be straightforwardly extended to those systems , in particular for achiral nanotubes , i.e. the armchair and zig - zag nanotubes considered here @xcite . for these particular systems with a uniform magnetic field along their axis , we find by using again a nearest - neighbours tight binding model , the following spectrum @xcite @xmath42 where for zig - zag nanotubes @xmath43and for armchair nanotubes @xmath44the roles of @xmath45 and @xmath46 are just exchanged from one case to the other .
using the expressions ( [ spectrumnano]-[armchair ] ) , one notices that - as for the case of rolled square lattices - the ground state of rolled honeycomb lattices changes also strongly in a magnetic field .
therefore , as for the case of rolled - square lattices , the static electric polarisability of carbon nanotubes are also expected to be strongly influenced by an axial magnetic field @xcite .
it is the purpose of the next section to analyse the structure of the dielectric susceptibility of cylindrical organic nanostructures in presence of an applied magnetic field .
moreover , the electronic spectrum of carbon nanotubes is also a periodic function of the magnetic flux , a behaviour which has been confirmed experimentally in @xcite .
very precise measurements of the polarisability can be done nowadays .
two recent examples are given in @xcite for multicomponent glasses and in @xcite for mesoscopic metallic rings .
the complicated electronic spectrum of cylindrical organic nanostructures - including carbon nanotubes - induced by an axial magnetic field causes important effects in the polarisability of these systems as we want to discuss in the following . for simplicity
let us focus on rolled square lattices only .
we give one example for carbon nanotubes in this section but , more informations can be found in @xcite . with a uniform dc electric field
@xmath47 , the tight binding model becomes @xmath48where @xmath49 is the dipole operator . in the following ,
two different orientations for the electric - field are considered .
the corresponding expression for the dipole operator are @xmath50 for longitudinal field , where @xmath13 is the lattice spacing and @xmath51 is the total length of the tube , and @xmath52 for transversal field , where @xmath53 is the radius of the cylinder ( fig .
the first case was studied in @xcite , the second in @xcite .
we are interested in the behaviour of the induced dipole moment as function of the magnetic flux @xmath54where , as usual , @xmath55 and @xmath56 is the chemical potential which is a function of the magnetic field .
we summarise in the following , the main results of the studies done in @xcite . in this case , the hamiltonian of the system is given by eqs .
( [ tbe ] ) and ( [ dipoleoperatorl ] ) . with this model
it becomes tedious to evaluate the polarisability ( [ dipole ] ) , for long chains .
instead of doing exact calculations we proceed by using a semi - classical approximation very powerful in this specific context @xcite .
we are interested in the linear regime where the potential induced by the electric field , @xmath57 , is slowly varying on the scale of the fermi wave length .
then , the thomas - fermi approximation may be applied @xcite .
we assume that it is meaningful to consider an energy depending on the position , @xmath58 , on the chain @xmath59 where @xmath60 is given by the expression ( [ spectrum ] ) .
next , within this semi - classical approximation , the polarisability is evaluated as follows .
@xmath61 at this stage , it should be noticed that the interaction which leads to an energy splitting at crossing points when the electric field is turned on , is neglected here .
therefore , a possible first order stark effect near those points is not taken into account with this approximation . to illustrate the relevance of this approximation
, we consider first a linear chain in the absence of a magnetic field . in this case , the quasi - classical energy at position @xmath58 is given by @xmath62 and the corresponding polarisability by @xmath63 figure 5 shows an example for a linear chain with 201 sites .
the exact polarisability is compared with the semi - classical one ( eq . [ tfchain ] ) as function of the electron filling for two different values of the temperature .
one can see that the deviations caused by the semi - classical approximation are less than @xmath64 or @xmath65 .
this justifies the use of this approximation for cylindrical systems in a magnetic field .
figure 6 shows the results for a cylinder with @xmath66 sites along the circumference and @xmath67 sites along the cylindrical axis .
the temperature is @xmath68 in units of @xmath69 ( around 1k ) and the density is 0.74 electrons per site . since the polarisability is symmetric with respect to the point @xmath70 , as mentioned before , only the regime @xmath71 is shown .
obviously there is a huge effect of the magnetic field on the polarisability .
these conclusions are based on a semi - classical approximation . in the next subsection we present full quantum calculations for the case where the electric field is perpendicular to the cylindrical axis .
therefore , we postpone a complete discussion of the results to the next subsection where they do not suffer from any approximation in the context of our simplified model .
the hamiltonian describing the system is now given by eqs .
( [ tbe ] ) and ( [ dipoleoperatort ] ) .
the advantage of this choice is to allow for a separation of variable : the motion of the electrons along the circumference and the cylindrical axis can be treated separately .
exact calculations can be done even for very long cylinders .
proceeding that way , the problem is reduced to a study of one - dimensional rings in applied electric and magnetic fields @xcite .
we show in fig . 7b the magnetic field dependence of the polarisability for a cylinder with @xmath72 and @xmath40 and very few electrons on it , @xmath73 ( this corresponds to a band filling of only @xmath74 ) . in this example
, the electric field is such that @xmath75 ( linear regime ) and the temperature is @xmath76 ( @xmath77 ) .
we choose this example because it shows very clearly the main behaviours of the dielectric response .
in particular we choose @xmath72 because this gives an illustration of the signature of the linear stark effect .
together with this example the energies of the ground state and first excited state are shown without electric field as function of the magnetic flux ( fig .
first of all the induced dipole moment as function of the magnetic field is periodic with a period of @xmath29 and is symmetric with respect to @xmath38 ; these symmetries are already apparent in the spectrum ( [ spectrum ] ) .
second , the induced dipole moment shows clearly two main characteristics : ( i ) we can notice the presence of small peaks at @xmath78 , ( ii ) the induced dipole moment is a discontinuous function showing several jumps separating plateau - like sections . note that there is a slight curvature in the whole spectrum , which is related to the persistent current induced by the magnetic field . as noticed before , a magnetic field induces crossing between the energies of the ground state and the first excited state ( figure 4 and 7a ) . as it can be seen in figures 7 , there is a one to one correspondence between level crossing at zero electric field and each kink in the induced dipole moment .
at each crossing , the ground state of the system changes .
these two states which are crossing respond differently to an applied electric field . both produce a quadratic stark effect but generally of different size .
this explains why the induced dipole moment is not a continuous function of the magnetic field .
near the crossing points , the response of the system will depend on whether or not there is a nonzero matrix element of the electric field between the two states involved .
if the matrix element vanishes , the picture described above is valid . on the contrary , if there is an interaction between those states , due to the degeneracy , the response will result in a linear ( instead of quadratic ) stark effect with peaks of the induced dipole moment .
more precisely , using the expression of the wave function ( [ wavefunction ] ) , the matrix elements of the dipole operator are calculated as @xmath79 with this equation , it is easy to see that a linear stark effect can occur only for a magnetic flux of @xmath80 .
these are precisely the values for which one obtains peaks in our first example ( fig .
it is worth noticing that large enough coulomb interactions can change drastically the selection rules ( [ selectionrules ] ) .
the same is true for distortions of a cylinder as it was shown in @xcite for elliptical cylinders .
appearance of a linear stark effect for different values of the magnetic flux could then give a way to quantify importance of electronic correlation effects or to characterise the geometry of the systems under consideration . at low enough temperature , one deals essentially with the spectroscopy of a few levels around the fermi level .
therefore it is not surprising that the induced dipole moment for a system which is well ordered , well oriented and of well defined size ( @xmath81 ) , depends strongly on the electronic density .
this is clearly seen from figs . 8a and 8b .
they are for the very same system considered before but for three different electron fillings @xmath82 and @xmath83 . at such a low density , the relevant mean level spacing , @xmath84 , behaves like @xmath85 .
figure 8a is for a temperature lower by one order of magnitude than @xmath84 , while fig .
8b is for a temperature higher by one order of magnitude than @xmath84 . with a typical value for @xmath69 ( @xmath86 )
one can estimate a temperature of @xmath87 for the spectrum 8a and @xmath88 for the spectrum 8b .
with these figures we want to emphasise the unique sensitivity of the proposed measurements and also that it is the necessary to work at very small temperatures in order to extract maximum informations .
the last examples in fig .
9 is an illustration showing that the same kind of behaviour is expected in carbon nanotubes .
it is for armchair ( fig .
9a ) and zig - zag ( fig .
9b ) nanotubes , when both are @xmath74 away from half - filling and at @xmath89 @xcite .
today it is possible to measure accurately very small variations in the real part of the dielectric function @xcite .
therefore , the dramatic magnetic field effects on the static polarisability of mesoscopic cylindrical systems , discussed here , could be measured and analysed .
several important informations about excited states could then be obtained .
first , the positions of the discontinuities and peaks should give informations about the energies of the excited states .
the nature of the response - linear or quadratic stark effect could be detected and therefore should give informations about the symmetry of the excited states .
the magnitude of the response should also give informations about the coupling constants .
finally , the different curvatures observed in the whole spectrum could give informations related to the persistent current - or ring current - induced by the magnetic field .
until now we were discussing cylindrical systems , such as carbon nanotubes , with a diameter in the mesoscopic range .
the diamagnetic current induced by the applied magnetic field is a ring current in a strict geometrical sense . in the following
, we will discussed quite different systems closer to those discussed by ehrenfest , pauling and london @xcite .
indeed , we consider finite graphene planes where the magnetic field induces a ring current running along the perimeter of these systems .
it is very close to the earlier studies where ring currents in benzene or naphthalene molecules were considered .
the mesoscopic graphene sheets are new materials of importance for which the magneto - polarisability @xcite could give very useful informations .
very recently a new type of magnetism similar to a spin glass state was found for activated carbon fibres prepared at heating temperature over 120023 c @xcite .
these materials are supposed to be well described by disordered networks of nanographite layers with a characteristic length of some tens of angstms @xcite .
it is worth noting that further increase of the heating temperature will drive the system towards complete graphitisation . on the other hand , recent theoretical works on graphite ribbons
have shown the existence of edge states whose wave functions are localised in the near vicinity of the edges of the ribbons and whose energies form a band with a small dispersion centred in the middle of the gap @xcite . therefore , these states can have important physical consequences since they make a relatively large contribution to the density of state at the fermi level .
for instance , they are expected to affect novel electrical and magnetic properties such as a paramagnetic behaviour at low temperature @xcite . in the light of these theoretical works ,
it was suggested by shibayama et al . that the edge states should be responsible for the spins observed in @xcite .
the spin glass behaviours could then be explained by invoking heisenberg like interactions between these effective spins since the disordered network of nanographite layers gives rise to a random exchange interaction network . in this context , we want to stress in this section that measurements of the magneto - polarisability may be a very direct and efficient way to detect possible edge - states in activated carbon fibres or any other amorphous carbon material . this new way of probing electronic structure of carbon materials could help to support the hypothesis made in @xcite .
moreover , we believe that this quantity could give a way to access characteristic length of graphitic regions in such compounds - quantity studied today using raman spectroscopy @xcite . for that purpose ,
we consider first rectangular finite planes of graphite ( fig .
these simplest systems show two different kinds of edges : zig - zag and armchair edges .
the one - band @xmath0 electrons are again described by a nearest - neighbours tight binding hamiltonian defined on the topologically equivalent brick - type lattice already used in @xcite , @xmath90 where the summation is over nearest - neighbour sites located at @xmath91 coordinates , @xmath92 and @xmath93 .
as found in references @xcite , our studies of ( [ tbgraphene ] ) show the existence of edge states for finite rectangles of graphite .
moreover , in agreement with the results of @xcite for ribbons , the wave functions of the edge states are shown to be localised preferentially on the zig - zag edges .
in particular the weight of these functions is negligible on the armchair edges .
this remark could be of importance for real materials leading to the question about the stability of such states in layers with disordered shape .
we postpone a discussion of this important question to a future publication . finally , the energies of these states are near the middle of the gap ; the closer it is to the middle , the more localised on the edges is the corresponding wave function . next we apply a uniform magnetic field @xmath94 perpendicular to the layer . in the tight binding model
we then proceed once more by using the london substitution @xcite , @xmath95 , where @xmath96 is the vector potential .
we write @xmath97where @xmath98 and @xmath99 are the appropriate line integral of the vector potential . since the magnetic field is uniform
, one must have ( cf fig .
10 ) , again in units of @xmath11 @xmath100 for calculations we choose , as in @xcite , the landau gauge .
@xmath101 it should be noticed that the elementary magnetic flux is defined here with respect to the size of a unit cell , namely a benzene ring , which requires an extremely large field except for the edge states where the relevant unit is given by the size of the whole layer - corresponding to a much more reasonable magnetic field of some tens tesla . in that sense ,
the edge - states define an effective rings .
the idea is quite simple : only the edge states will react to small values of an applied magnetic field such that the corresponding characteristic length of the cyclotron radius is of the same order than the size of the layer .
subsequently , the signature of the edge states should be seen in the polarisability as function of magnetic field .
we then add a static uniform electric field in the plane of the layer and calculate the magneto - polarisability for different electron - fillings . in this case
, the dipole operator takes the following general expression @xmath102 where the site vector @xmath103 is defined with respect to an arbitrary origin .
we choose a system with a total of 76 benzene rings with 10 along the zig - zag edges .
we work at small temperature @xmath104 which could be estimated to @xmath89 .
the particular example studied here shows only two nearly degenerate edge states at zero energy .
the edge states will be populated at half - filling but since the temperature is low compared to the energy gap , they are not occupied if one reduces the electron number by one .
figure ( 11 ) shows the magneto polarisability for these two cases .
we see very clearly a quadratic dependence of the polarisability as function of the magnetic field for the half - filled case only . for the other case ,
no sensible effect are seen .
we can then conclude that the edge states are responsible for this quadratic behaviour in accordance with the argument given above . to conclude this section
, it should be stressed that the relative variations of the polarisability in fig .
11 for the case with occupied edge states can be detected nowadays .
therefore , we believe it is possible to identify the existence of edge states in activated carbon fibres , or other amorphous carbon materials , using measurements of the magneto - polarisability . moreover , we believe that a careful analysis of the quadratic behaviour shown in fig .
11 can help to gain important informations about these disordered materials such as the effective sizes of the graphite - like regions for instance .
so far we did nt consider coulomb interactions .
instead , we were assuming that they were included in the effective one - electron parameters of our tight - binding models .
however , for quantitative results of the polarisation , it is needed to include explicitely those interactions .
it is well known that the absolute value of the polarisation of organic compounds decreases strongly by including interaction corrections @xcite .
we will not attempt such studies here which will be the subject of a subsequent publication .
instead , we will consider several strongly interacting systems , rings or small cylinders , in the extreme limit of infinite on - site interactions , our purpose being to point out that coulomb interactions can produce not only quantitative changes but , also , important qualitative changes in the behaviour of the polarisability . in the early nineties ,
several authors have studied the behaviour of persistent currents in mesoscopic hubbard rings threaded by a magnetic field @xcite .
one of the most prominent result of these studies was obtained in the limit of infinite interaction , @xmath105 . in this limit
, kusmartsev pointed out that , in the case of spin @xmath38 fermions , the persistent current has a surprising short period given by @xmath106 ( in unit of @xmath11 ) , where @xmath107 is the number of electrons @xcite .
he called this intriguing new phenomena , fractional aharonov - bohm effect .
it is this result which we shortly describe here .
the tight binding hubbard model for a ring with @xmath6 sites is of the form @xmath108 where @xmath109 is the usual electron number operator .
the magnetic flux @xmath10 is given by the expression ( [ flux ] ) .
an advantage of the hubbard ring is to be one of the strongly - correlated - electron system that is solvable by using the bethe ansatz @xcite . in the presence of a magnetic field the form of the wave function
does nt change @xmath110\exp(i\sum_{j=1}^{n_e}k_{p_j}x_{q_j})\ ] ] where @xmath111 and @xmath112 are two permutations of @xmath113 .
the coefficients @xmath114 $ ] as well as @xmath115 are determined from bethe s equations which in a magnetic field are modified by the addition of the flux phase @xmath10 @xcite .
@xmath116 @xmath117@xmath107 is the number of charges i.e. the number of electrons , and @xmath118 is the number of down - spins .
the eigenstates of a ring are characterised by the momenta @xmath119 of the charges and the rapidities @xmath120 of spin waves .
the energy of the system is given by @xmath121 in the limit of infinite interaction @xmath105 bethe s equations are greatly simplified to give the following results @xmath122 where the quantum numbers @xmath123 and @xmath124 are connected with charge and spin degrees of freedom respectively .
they are either integers or half odd integers , depending on the parities of the numbers of down- and up - spin electrons : @xmath125 the equation ( [ uinfinity ] ) is identical to the one describing a set of non - interacting spinless fermions on a ring threaded by a flux @xmath126 .
the second term of this expression is the effective magnetic flux created by the spin excitations .
it comes from the fact that there is an additional phase shift between the momenta of the different electrons due to the repulsive interaction . because of this shift
, the periodicity of the aharonov - bohm flux takes a fractional value at the infinite interaction limit .
it is easy to check this statement by using eqs .
( [ energybethe ] ) and ( [ uinfinity ] ) and by appropriately choosing the quantum numbers ( [ quantumnumbers ] ) in order to minimise the energy of the system .
one example is shown in fig . 12a for a ring with 8 sites and 2 electrons , one up and one down .
another example is shown in fig .
12b for the very same ring but for 5 electrons , 3 up and 2 down .
the fractional aharonov - bohm effect is clearly seen in these examples .
it is important to stress once more that the spin plays an essential role in this phenomena .
this can be seen in ( [ uinfinity ] ) where the quantum numbers associated to the spin degree of freedom appear as responsible of the fractional ab effect . without this term
we recover the known result for free fermion rings ( [ eigenvaluering ] ) .
in particular , for spinless fermions , the periodicity with the flux is 1 . in the case of infinite interaction or spinless fermions ,
we can use the following argument to obtain the correct flux periodicity and which stresses the important role played by the spin .
let us take a particular configuration with @xmath127 electrons , @xmath128 .
one may ask the question of how many hops of the particles in the same direction should take place in order to recover the starting configuration . in mathematical terms
we may write @xmath129 where @xmath130 is a phase term cause by the pauli s principle . @xmath131 are general coordinates of the electron @xmath132 taking account of the spin when necessary .
@xmath133 are the operator of translation by one site in the clockwise direction ( for instance ) of the electron @xmath132 .
@xmath134 is the number of one - site - hopping of the particle @xmath132 .
the answer to this question depends on whether or not we consider the spin degree of freedom . in the cases we are interested in
, the fermions are locked together and form a kind of dynamical wigner crystal . by moving on the lattice , an electron will push the others in the same direction . for the case of spinless fermions , since all particles are indistinguishable , it is easy to see that @xmath135 to get back to the initial configuration .
the same is true for an assembly of free spin @xmath38 fermions . on the other hand ,
one come back to the same configuration by interchanging the particles which , due to the pauli s principle , give the following statistical flux : @xmath136 .
the immediate conclusions are , first , that the periodicity of the energy with the flux will be 1 ( as for the free electron case ) and , second , there will be a parity effect which discriminates between the cases with even and odd number of particles . indeed , it was shown by kusmartsev @xcite ( and references therein ) , that a system of interacting spinless fermions on a ring behaves as a paramagnet for even number of particles and as a diamagnet for an odd number of particles .
this property is sometimes known as _
leggett s conjecture _ @xcite . for the case of spin
@xmath38 fermions , the conclusions are completely different .
indeed , one easily sees that @xmath137 and @xmath138 .
each particle should go round the ring and therefore , the statistical flux is just unity .
consequently , the effective charge of the system will be @xmath139 and the periodicity of the energy will be @xmath140 - this is the fractional aharonov - bohm effect discovered by kusmartsev @xcite .
moreover , there is no parity effect : indeed , it is known that the system is always diamagnetic for infinite interactions @xcite . on the basis of these analysis ,
let us now consider strongly - interacting cylindrical systems in some limiting cases where conclusions could be reach following the same argumentation .
let us first consider a stack of isolated rings with only one electron per ring .
the rings are interacting following the model @xmath141 where there is a nearest - neighbour interaction between two electrons when they are on top of each other .
this hamiltonian could help to describe stack of large aromatic molecules with only one electron in each in the lumo .
it could also be an effective model to describe interacting edge - states in stack of finite graphene planes .
this model was already used in @xcite in the continuum limit , where an interesting new parity effect was pointed out to occur in the dielectric response of one - dimensional stack of quantum rings . here
, we are only interested by the periodicity of the ground state energy with the magnetic flux in the limit of infinite interaction . once again , starting from one particular electronic configuration , we are asking the same question about the number of particle hopping in the same direction one should do to come back to it .
it is then possible to write down a similar expression than ( [ periodicity ] ) with @xmath137 and @xmath138 where @xmath127 is the number of rings and @xmath6 , the number of sites in each ring .
consequently , one can conclude that the periodicity is @xmath140 and that there is no parity effect .
the label of the ring acts as a spin degree of freedom .
it is worth mentioning that it exists a rigorous mathematical proof of these properties @xcite .
last , we are considering a model of interacting cylinders @xmath142 with again @xmath143 .
again we are asking the same question about the periodicity formulating it in terms of equation ( [ periodicity ] ) .
one has to distinguish between the case of spin @xmath38 and spinless fermions .
* for spinless fermions , for a non vanishing @xmath144 , one has @xmath135 and @xmath138 ; the periodicity is one and there is no parity effect . * for spin @xmath38 fermions , whatever is the value of @xmath144 , one has @xmath137 and @xmath138 ; the periodicity is @xmath140 and there is no parity effect . in conclusion , there is an abrupt transition from a fractional aharonov - bohm state to an usual aharonov - bohm state going from @xmath145 to @xmath146 in the case of spinless fermions ; on the contrary , there is no such transition for spin @xmath38 fermions .
the main purpose of this review is to point out the strong magnetic field dependence of the static polarisability of nanoscopic cylindrical organic materials when the magnetic field is parallel to the tube axis .
this was done by studying two kinds of system .
on one hand we have considered real cylinders in the form of rolled square lattices but also , rolled honeycomb lattices , more specifically , achiral carbon nanotubes .
they were studied with both , semiclassical approximation @xcite and exact diagonalisation @xcite .
for all these cases , the polarisability was shown to present very complex structures as function of the magnetic field in which one can identify two different characteristics ( cf fig .
7 ) : ( i ) the polarisability is a non - continuous function with sudden jumps separating plateau - like regions ( ii ) additionally , small peaks may appear for special values of the magnetic field in place of jumps .
a full understanding of these complicated behaviours was given by following the behaviour of the ground state by increasing the magnetic field : due to the aharonov - bohm effect , many changes of ground state occur and for some values of the magnetic field accidental degeneracies happen where the ground state becomes two fold degenerate .
a one to one correspondence is found between the accidents in the polarisability and the accidental degeneracies of the ground state .
each plateau - like region of the polarisability corresponds to a quadratic stark effect with a coefficient proper to the corresponding magnetic - field induced ground state .
each peak corresponds to a linear stark effect appearing at accidental degeneracies when there is direct coupling between the two states involved ( cf fig .
on the other hand we have studied finite plane of graphene which support edge states .
the wave functions of these states have the peculiarity to be mostly localised on the perimeter of the plane defining kind of effective rings .
the presence of these states in real materials is a important question since they are able to influence new physical properties @xcite and , we have shown that it is possible to use measurements of the static polarisability in an applied perpendicular magnetic field to detect possible edge states in amorphous carbon materials such as carbon fibres .
moreover we believe that a careful analysis of the quadratic behaviour we have pointed out in section iv ( fig .
11 ) , should give a way to determine the characteristic length of the graphitic region in these disordered systems .
in a last part of this review , we have briefly described results of some strongly interacting models in the limit of infinite interactions .
the main purpose of this section was to pointed out that the interaction can result in new phenomena such as the fractional aharonov - bohm effect which appears for hubbard rings with spin @xmath38 fermions but also in stacks of uncouple rings with one electron per ring .
an explicit treatment of the electron - electron interactions is therefore needed and , this is now our main direction of investigation .
p. ehrenfest , physica * 5 * , 388 ( 1925 ) ; z. phys .
* 58 * , 719 ( 1929 ) .
l. pauling , j. chem .
* 4 * , 673 ( 1936 ) . f. london , j. phys .
radium * 8 * , 397 ( 1937 ) .
dresselhaus , g. dresselhaus and p. eklund , `` science of fullerenes and carbon nanotubes '' , academic press ( 1996 ) .
a. bachtold , c. strunk , j.p .
salvetat , j.m .
bonard , l. forro , t. nussbaumer and c. schnenberger , nature 397 , 673 ( 1999 ) .
p. strehlow , c. enss and s. hunklinger , phys .
lett . * 80 * , 5361 ( 1998 ) ; p. strehlow , m. wohlfahrt , a.g.m .
jansen , r. haueisen , g. weiss , c. ens and s. hunklinger , phys .
lett . * 84 * , 1938 ( 2000 ) .
p. fulde and a.a .
ovchinnikov , eur .
j. * b 17 * , 623 ( 2000 ) .
s. pleutin and a.a .
ovchinnikov , eur .
j. * b 23 * , 521 ( 2001 ) .
s. pleutin and a.a .
ovchinnikov , carbon * 40 * , 129 ( 2002 ) .
k.h . ahn and p. fulde , phys . rev . *
b 62 * , r4813 ( 2000 ) .
efetov , phys . rev
. lett . * 76 * , 1908 ( 1996 ) .
y. noat , b. reulet and h. bouchiat , europhys . lett .
* 36 * , 701 ( 1996 ) .
m. blanter and a.d .
mirlin , phys .
b 57 * , 4566 ( 1998 ) .
r. deblock , y. noat , h. bouchiat , b. reulet and d. mailly , phys .
lett . * 84 * , 5379 ( 2000 ) .
u. eckern and p. schwab , j. low . temp .
phys . * 126 * , 1291 ( 2002 ) . k. richter , d. ullmo and r.a .
jalabert , phys . rep . * 276 * , 1 ( 1996 ) .
n. bayers and c.n yang , phys .
lett . * 46 * , 7 ( 1961 ) .
r. saito , g. dresselhaus and m.s .
dresselhaus , `` physical properties of carbon nanotubes '' , imperial college press ( 1998 ) n.w . ashcroft and n.d .
mermin , _ solid state physics _ ( saunders college publishing , new york , 1976 ) .
y. shibayama , h. sato , t. enoki and m. endo , phys .
* 84 * , 1744 ( 2000 ) .
m. fujita , k. wakabayashi , k. nakada and k. kusakabe , j. phys .
. jpn . * 65 * , 1920 ( 1996 ) .
k. wakabayashi , m. fujita , h. ajiki and m. sigrist , phys .
b 59 * 8271 ( 1999 ) .
a.c . ferrari and j. robertson , phys . rev . *
b 61 * , 14095 ( 2000 ) .
benedict , s.g .
louie and m.l .
cohen , phys . rev .
* b 52 * , 8541 ( 1995 ) .
kusmartsev , j. phys . condens .
matter * 3 * , 3199 ( 1991 ) .
b.s shastry and b. sutherland , phys .
lett . * 65 * , 243 ( 1990 ) ; a.a .
zvyagin , fiz .
nizk . temp .
* 16 * , 1299 ( 1990 ) [ sov . j. low
* 16 * , 745 ( 1990 ) ] ; n. yu and m. fowler , phys . rev .
* b 45 * , 11 795 ( 1992 ) .
lieb and f.y .
wu , phys .
* 20 * , 1445 ( 1968 ) .
leggett , in _ granular nanoelectronics _ , edited by d.k .
ferry et al .
( plenum press , new york , 1991 ) , p. 297 ; d. loss , phys . rev .
69 * , 343 ( 1992 ) .
kusmartsev , j.f .
weisz , r. kishore and m. takahashi , phys .
rev * b 49 * , 16 234 ( 1994 ) .
a. bernet , unpublished . | we review recent results on the behaviour of the dielectric function of cylindrical nano - organic materials at very low frequencies in a magnetic field . for cylindrical structures - such as carbon nanotubes - the polarisability is shown to be a discontinuous function of a longitudinal magnetic field where plateau - like regions are separated by sudden jumps or peaks .
a relation is pointed out between each discontinuity in the polarisability and the cross - over between ground and first excited states induced by the magnetic field .
this one to one correspondence suggests to use measurements of the dielectric function in an applied magnetic field in order to obtain informations about the electronic structures of cylindrical nanostructures .
in addition , it is shown , by studying finite graphene layers , that the measurement of the polarisability in a magnetic field could be a powerful way for detecting possible edge - states in amorphous carbon materials such as activated carbon fibres .
finally , the importance of the electron - electron interaction is emphasised by discussing examples of strongly interacting electrons on rings or cylinders , in the limit of infinite interaction . |
the dynamics of cold atomic vapours trapped in a periodic external potential , as created by means of a far - detuned standing wave of light , has become an active area of research .
studies of ultracold atoms in such a confinement have revealed a number of behaviours which are well known in solid state physics : wannier - stark ladders [ 1 , 2 ] , bloch oscillations [ 3 , 4 ] and landau - zener tunneling [ 5 ] .
more recently , the observation of interference between macroscopic quantum states of bose - einstein - condensed vapours of @xmath1rb atoms confined in a vertical array of optical traps has been reported [ 6 ] .
interference between the condensates in different lattice wells manifests itself in falling drops , which are emitted with a period determined by the bloch oscillations induced by the gravitational field and are interpreted as coherent matter - wave pulses .
the properties of a bose - einstein condensate ( bec ) in a regular array of optical wells have been studied theoretically and numerically by a number of authors [ 7 - 13 ] . especially relevant for our present purposes is the work on an interacting bec in an infinitely extended lattice potential by berg - srensen and mlmer [ 7 ] , who demonstrated band structure and bloch oscillations by solving numerically the bogolubov - de gennes equations for the excitations and the time - dependent gross - pitaevskii equation for motion under a constant applied force .
they also gave analytic results for the excitation spectrum in the limit where the periodic potential can be treated as a perturbation . from solid state theory
a tight - binding ( tb ) approach offers the possibility of an analytic study of the same phenomena in the opposite limit , when the lattice potential is so strong compared with all other energy scales as to allow the use of on - site orbitals for the approximate construction of the bloch states .
in fact , the proper theoretical tools are the so - called wannier functions , which are defined through the fourier transform of the bloch eigenfunctions in the momentum representation . in the particular case of one - dimensional ( 1d ) motion ,
the energy bands are non - degenerate and the states in any given one of them are described from a single wannier function , by repeating it at each lattice site with the appropriate phase relationship between different sites .
the tb scheme may then be realized by replacing the wannier function by an on - site `` atomic '' orbital , under suitable limiting conditions which may be valid for the low - lying energy bands . while the wannier functions for motion in a generic periodic potential are usually not easy to calculate , it turns out that for the specific problem of 1d periodicity posed by the confinement of a bec in an array of optical wells one can take advantage of the results contained in a classic paper by slater [ 14 ] on a soluble problem in energy bands for electrons .
slater s work also discusses the conditions under which the wannier functions for the low - lying energy bands can be accurately approximated by on - site orbitals , which in this case are the harmonic - oscillator wave functions associated with the parabola describing the bottom of each well .
the plan of the paper is as follows . in section 2
we first model a dilute bec confined in a periodic optical potential as a 1d dynamical system via inclusion of the transverse size into a renormalized contact interaction .
we then formulate the problem of 1d band structure in terms of wannier functions , the only conceptual difference from the analogous problem of 1d motion of electrons in a periodic potential being that for condensed bosons there is no gap between the condensate ground state and the excitation spectrum . in the same section
we also formulate the problem of bloch oscillations of the condensate in a single band in terms of the fourier transform of the corresponding wannier function .
the reduction of the general formalism to the tb approximation is effected in section 3 , after recalling slater s conditions for its validity and extending them to an interacting bec .
the tb scheme displays in an explicit manner how the energy levels of a bec in a single well are shifted and spread into bands by the periodic replica of the well potential .
we give special attention to ( i ) the lowest - lying energy band of the bec , which contains its wave - like phase excitations , and ( ii ) the connection that the tb approximation exhibits between the excitations in the next two higher bands and the dipole ( `` sloshing '' ) and quadrupole excitations in an effectively 1d harmonic well .
the discussion suggests how excitations in these two bands may be observed .
we conclude this section by simple numerical illustrations of the dispersion relation in the lowest band and of the current induced by a constant external force , for values of the system parameters which are relevant to the experiments of anderson and kasevich [ 6 ] .
finally , section 4 ends the paper with some concluding remarks .
we consider a dilute vapour of atoms of mass @xmath2 in a bose - condensed state at zero temperature , which interact via elastic collisions described by an s - wave scattering length @xmath3 and are confined in the optical lattice created by a laser field .
the lattice can be modelled as a transverse gaussian potential times a periodic potential along the @xmath4 direction , namely the confining potential is @xmath5 here @xmath6 is the well depth , scaling linearly with the intensity of the laser beam , @xmath7 is the transverse size of the beam and @xmath8 its wavenumber .
the lattice period is @xmath9 . in this work
we reduce the above problem in cylindrical symmetry to an effective 1d problem . to this end
we freeze the transverse motions of the bec and renormalize the interactions by a method proposed by jackson _ et al . _
[ 15 ] to treat an elongated bec , in which the coherence length as introduced by baym and pethick [ 16 ] is much larger than the transverse size .
this reduction will be inapplicable , however , to high - energy excitations , when the excitation energy becomes comparable with the transverse confinement energy and transverse motions can therefore be induced . under the conditions indicated just above , the transverse confinement modifies the mean - field interactions by a factor depending on the axial density @xmath10 , the factor being @xmath11 where @xmath12 is the transverse part of the wave function in the @xmath13 plane . using a harmonic approximation for the radial part of the optical potential , we obtain in the present problem an effective scattering length per unit area given by @xmath14 where @xmath15 is the number of atoms per lattice well and @xmath16 .
notice that the energy @xmath17 differs by a factor eight from the recoil energy @xmath18 .
a similar renormalization of @xmath6 in eq .
( [ perpo ] ) is negligible , since @xmath19 .
the effective 1d hamiltonian @xmath20 that we treat thus is @xmath21 where @xmath22 is the field operator , @xmath23 and @xmath24 , with @xmath25 the chemical potential and @xmath26 . in the following all energies
are referred to the bottom of the lattice potential .
let us summarily go through the standard bogolubov procedure [ 17 ] for treating the boson hamiltonian ( [ hamiltonian ] ) .
we make the ansatz @xmath27 , which separates the coherent condensate function @xmath28 from the fluctuating part @xmath29 in the field operator . upon substituting this ansatz in eq .
( [ hamiltonian ] ) and requiring that linear terms in @xmath29 vanish , one obtains the gross - pitaevskii equation for @xmath28 : @xmath30\phi(z)=0\ ; .\ ] ] we are taking @xmath28 as real without loss of generality , with normalization to unity over the lattice period . to quadratic terms
the hamiltonian in the fluctuation operators is diagonalized by the canonical transformation @xmath31\ ; , \nonumber\\ \tilde{\psi}^\dagger(z)&=&\sum_\alpha\left[u^*_\alpha(z)a^\dagger_\alpha- v^{*}_\alpha(z)a_\alpha\right]\ ; .\label{bogo}\end{aligned}\ ] ] in eq .
( [ bogo ] ) the suffix @xmath32 denotes an appropriate set of quantum numbers and the functions @xmath33 satisfy the bogolubov - de gennes equations , @xmath34 here , @xmath35 , @xmath36 and the @xmath37 s are the ( real ) excitation energies that we aim to determine .
we explicitly point out that with this form of @xmath38 the interactions enter the determination of the excited - state wave functions only through the density distribution of the condensate ground state .
the same assumption has been made in dealing with the excitations of condensates in harmonic traps [ 18 ] and has led to good agreement with experiment for the eigenfrequencies of the quadrupolar surface modes . for a bec in a 1d periodic potential the quantum numbers @xmath32 are the band index @xmath39 and the lattice quasi - momentum @xmath40 . from symmetry under time reversal
we have that , if @xmath41 is a solution of eqs .
( [ bogoeq ] ) corresponding to energy @xmath42 , then @xmath43 is a solution corresponding to energy @xmath44 [ 19,20 ] .
we may for convenience call these the particle - like and the hole - like solution , respectively , and limit ourselves to calculate spectrum at positive energy .
as for the homogeneous bose fluid we expect gaplessness , since in its ground state the whole condensate is in a single energy level placed at the chemical potential . therefore , the excitations of lowest energy ( referred to the chemical potential ) are phonons , i.e. they have a linear dispersion relation .
it is convenient to introduce solutions @xmath45 which have definite symmetry under time reversal [ 7 ] , by setting @xmath46/\sqrt{2}$ ] . from equations ( [ bogoeq ] ) we have @xmath47 and @xmath48 where @xmath49
hence , @xmath50 this equation shows that the functions @xmath51 and @xmath52 are the right and left eigenvectors of the non - hermitean product operator @xmath53 .
the ground state of the condensate can be taken at @xmath54 in the lowest band ( @xmath55 , say ) and special care is needed in handling the functions @xmath56 , in order to ensure gaplessness .
this point is discussed for a periodic condensate in appendix a , following the arguments given in refs . [ 19 ] and [ 20 ] .
here we only need to note that an evaluation of the functions @xmath52 is not necessary for the determination of the excitation energy eigenvalues .
these are calculated by solving eq .
( [ bogoeqe2 ] ) for @xmath51 , which reads @xmath57 this equation applies also to the state @xmath58 at zero energy ( relative to the chemical potential ) .
of course , the function @xmath58 is simply proportional to the function @xmath28 introduced in eq .
( [ gpe ] ) .
the solutions of eq .
( [ bogoeqe2 + ] ) must have the bloch symmetry , which imposes definite @xmath40-dependent phase relationships between different wells .
it is well known from solid state theory [ 21 ] that any bloch orbital can be expressed as a superposition of the so - called wannier functions centred on the lattice sites .
we thus write @xmath59 where the index @xmath60 runs over all @xmath61 sites of the lattice .
of course , the wannier functions @xmath62 have the same parity under time reversal as the corresponding bloch orbitals @xmath45 . we may also recall that the wannier functions form a complete orthonormal set , i.e. they satisfy the relation @xmath63 .
orthogonality between different wells implies oscillatory tails , falling off more rapidly as overlap decreases .
( [ wannier ] ) expresses a fourier - transform relationship between bloch orbitals and wannier functions , embodying the bloch translational symmetry .
the @xmath40-dependent bloch orbitals emphasize the extended nature of the lattice states and the periodicity of their energy eigenvalues in momentum space .
the representation in terms of wannier functions emphasizes instead that the lattice is a periodic assembly of identical wells .
the bloch symmetry of the fourier series in eq .
( [ wannier ] ) is ensured by the fact that the fourier coefficients do not involve a separate dependence on position @xmath4 and on lattice site index @xmath60 . in the case of a bec in a lattice ,
a general consequence of the wannier - function description is immediately evident from eq .
( [ wannier ] ) .
since the ground state is at the chemical potential and has been taken to lie at @xmath54 in the bottom of the lowest band , the excitations in this band are wave - like modulations of the condensate phase and their energy vanishes linearly in the limit @xmath64 .
this fact can be traced back to the existence of a goldstone phonon - a collective mode without restoring force [ 19 ] resulting from spontaneous symmetry breaking ( see the discussion in appendix a ) . similarly , for the higher excitation bands
( [ wannier ] ) shows that each of them is a superposition of localized excitations ( described by @xmath65 ) , which is constructed by imposing definite phase relationships between the various sites as dictated by the translational lattice symmetry .
this theorem is valid , of course , within the bogolubov approach and has been derived through the reduction to one - dimensionality . in practice ,
the calculation of wannier functions is in general not an easy task . in the limit
in which the wannier functions become almost confined inside each lattice well ( which may be true for low lying states ) , they can be expanded in a limited number of `` atomic '' orbitals .
we shall discuss in the next section the conditions under which one may replace the wannier functions for a bec in a periodic optical potential by single - well orbitals and thereby obtain explicit solutions in the tb limit .
we proceed to express through the wannier functions the current carried by a periodic bose - condensed system subject to a constant force @xmath66 in the semiclassical regime .
we restrict ourselves to bloch oscillations in the lowest band , described by the wannier function @xmath67 . through eq .
( [ wannier ] ) taken at @xmath54 , this function is immediately related to the ground state @xmath28 of the condensate .
the quantity of interest is the expectation value @xmath68 of the momentum operator @xmath69 in the state @xmath70 , where @xmath71 .
we introduce the fourier transform @xmath72 of @xmath67 , @xmath73 hence , @xmath74\ ] ] and a straightforward calculation yields @xmath75\over \sum_l \int_{-\infty}^{\infty } dk\ ; holds independently of the confining periodic potential and of the interaction between the particles .
it is easily seen from eq .
( [ pmedio ] ) that ( i ) @xmath68 is a periodic function of time with the bloch period @xmath76 and ( ii ) @xmath68 vanishes at the brillouin zone edges , i.e. at @xmath77 or @xmath78 .
the aim of the tb approximation is to relate the properties of the periodic system in a lattice potential to the solution of the single ( isolated ) well problem .
the results of the method are quantitative as long as the single - well solutions at adjacent sites do not overlap much with each other .
namely , the validity of the tb approximation is limited by the condition that the band - width be smaller than the height of the lattice potential .
slater [ 14 ] has studied the wannier functions of the `` mathieu problem '' defined by the schrdinger equation for non - interacting electrons moving in the periodic potential @xmath26 of present interest .
he showed that for the first five low - lying energy bands the wannier functions are well approximated by hermite polynomials under the condition @xmath79 , relaxing to @xmath80 if only the lowest energy band is of interest .
slater s argument is summarized in appendix b and extended there to an interacting bec .
in essence the mean - field interactions may be viewed in the present context as effecting a renormalization in the height of the external potential , thereby shifting slater s thresholds by the amount @xmath81 if the threshold conditions are satisfied , the wannier functions @xmath65 at positive energy could be approximately replaced by the eigenfunctions of the harmonic approximation to the single - well potential @xmath82 with @xmath83 $ ] and @xmath84 . more generally , the evaluation of the energy bands for a periodic condensate in the tb approximation would require a preliminary solution of the on - site problem . to this end , we separate the on - site parts @xmath85 out of the operators @xmath86 entering eq .
( [ bogoeqe2 ] ) by writing @xmath87 .
the detailed definitions of these operators are given in appendix c. the on - site operators @xmath85 determine the single - well orbitals @xmath88 corresponding to single - well energy eigenvalues @xmath89 , according to @xmath90 again , the condensate in a single well lies at the chemical potential @xmath91 ( see the definitions of @xmath85 in appendix c ) .
conjugation between fluctuations in the particle number and in the phase must also be taken care of , as shown in refs .
[ 19 ] and [ 20 ] .
we define the overlap integrals @xmath92 and the interaction integrals @xmath93\varphi^{}_{n+}(z - ld)$ ] ( see the definitions of the @xmath94 operators given in appendix c ) . by exploiting eqs .
( [ bogoeqs ] ) one easily finds from eq .
( [ bogoeqe2 ] ) the dispersion relation @xmath95 an adiabatic argument regarding the switching - on of the ( spatial - symmetry conserving ) mean field interactions shows that the product @xmath96 is even under space inversion . both series in eq .
( [ bandgen ] ) are therefore real quantities .
( [ bandgen ] ) displays the two main changes induced by lattice periodicity on the single - well energy levels , namely ( i ) a shift of each level relative to the single - well value @xmath89 and ( ii ) the broadening of the level into a band .
in particular , for the lowest band the shift corresponds to a renormalization of the chemical potential from the single - well value @xmath91 to the value @xmath25 , which has to be evaluated self - consistently from the condition @xmath97 .
the resulting gaplessness introduces in eq .
( [ bandgen ] ) for @xmath55 the linear , acoustic - phonon - like dispersion of the excitation energies in the long wavelength limit @xmath64 .
we can now see how the excitations in the bands with @xmath98 and @xmath99 may be related to single well excitations .
we make use of the present replacement of wannier functions by on - site orbitals , which are approximately described by the eigenfunctions of the 1d harmonic oscillator .
evidently , this allows us to identify the on - site excitation leading to the @xmath98 band with the dipolar sloshing mode of the condensed particles within the well .
the construction of bloch orbitals from these localized sloshing motions simply amounts to fixing , for each value of the wavenumber @xmath40 , their relative phases according to the lattice translational symmetry ( see the discussion already given near the end of section 2.2 ) . similarly , the excitations in the @xmath99 band may be viewed as resulting from localized , single - well quadrupolar modes with relative phases depending on @xmath40 as imposed by lattice symmetry .
of course , the approximate validity of such a visualization of the collective low - lying modes of a periodic bec is subject not only to conditions relating to effective one - dimensionality in a mean - field approach , but also to the restrictions associated with the tb scheme .
the above discussion also suggests how these collective band modes may become accesible to experimental or computational observation
. clearly , at @xmath54 the dipolar mode can be excited by a rigid shaking of the lattice and the quadrupolar one by a modulation of the width of all the wells .
excitation of dipolar and quadrupolar band modes at finite @xmath40 would require similar probes at given wavelengths .
we conclude this section by presenting some illustrative numerical results for the lowest band in the tb approximation . according to eq .
( [ slaterineq ] ) , in the case of repulsive interactions the inequality @xmath100 must be satisfied for the appropriate wannier function to be approximated by the gaussian function @xmath101 the width @xmath102 of this gaussian is determined in a first approximation by the curvature of the bottom of the single well .
a more accurate estimate of @xmath102 , which takes account of the interactions between the particles in the bec , is given in appendix c. in fact , for the calculation of the dispersion relation it is more convenient to revert back to eq .
( [ bogoeqe2 + ] ) and directly evaluate the integrals @xmath103\varphi^{}_{0+}(z - ld)\ ] ] and @xmath104 .
this procedure has the advantage of involving only the known symmetric function @xmath105 given in eq .
( [ gauss ] ) . since @xmath106
, we obtain @xmath107\;\ ] ] where @xmath108 .
we recall that the energies in eq .
( [ bandlow ] ) are referred to the bottom of the lattice potential .
we have calculated the integrals entering eq .
( [ bandlow ] ) with the wave function in eq .
( [ gauss ] ) for @xmath109 , which corresponds to a value of @xmath110 , in the case @xmath111 , @xmath112 and @xmath113 .
these values are relevant to the experiment of anderson and kasevich [ 6 ] .
the results are @xmath114 and @xmath115 , yielding the value @xmath116 for the speed of phonon - like excitations .
figure 1 shows the corresponding dispersion curve over the whole brillouin zone .
( [ gauss ] ) in eqs .
( [ ft ] ) and ( [ pmedio ] ) we find the following expression for the current induced by a constant external force in states of the lowest band , @xmath117 \over 1 + 2\sum_{l=1}^{\infty } \cos ( lqd)\exp \left[-\left ( ld/2\sigma\right)^2\right]}\ ; .\ ] ] in practice , in the tb limit the leading term in the numerator is dominant and the denominator can be replaced by unity .
namely , @xmath118\ ; .\ ] ] figure 2 shows the behaviour of @xmath119 over the brillouin zone for various values of the parameter @xmath120 in the case @xmath121 , @xmath112 and @xmath113 .
in summary , and as already discussed in the work of berg - srensen and mlmer [ 7 ] , the properties which derive from lattice symmetry for the wave functions and the energy eigenvalues of electrons moving in a periodic 1d potential also hold for a bose - condensed system . however , bose statistics and spontaneous symmetry breaking imply that there is no gap between the condensate ground state and the excitation spectrum .
this latter property can be met in the energy spectra of electrons in semiconductors only as a consequence of accidental degeneracies induced by specific space symmetries , a well known instance being provided by graphite [ 22 ] . for a periodic bose condensate the gaplessness property , together with the quadratic structure of the energies as expressed in eqs .
( [ bogoeqe2 ] ) ( see section 2.1 ) , yields on general grounds the linear dispersion relation @xmath122 for the lowest energy band .
we have seen that the specific form of the optical potential acting on a bec from a laser field admits two important simplifications in the band - structure problem .
firstly , the problem for the low - lying bands can be mapped into a one - dimensional one by accounting for the transverse cross - section of the bec through a renormalization of the s - wave scattering length .
this mapping has been shown to be useful in relation to the experiments performed by anderson and kasevich [ 12 ] .
secondly , the 1d problem away from the weak - confinement limit is most naturally formulated in terms of the wannier functions , which have already been studied for the mathieu problem by slater [ 14 ] .
the use of wannier functions , as demonstrated in section 2.2 , provides an exact formulation of the lattice problem and immediately shows that ( i ) the lowest excitations of a condensate in a periodic potential are wave - like phase modulations , and ( ii ) the higher bands arise from the superposition of localized excitations with definite phase relationships .
an analytic study of semiclassical transport in a periodic condensate subject to a constant external force becomes straightforward with the use of wannier functions .
the expectation value of the momentum operator is periodic with the bloch period and vanishes at the brillouin zone edges , as shown in section 2.3 .
the work of slater [ 14 ] allows one to specify the conditions under which the wannier functions for the low - lying bands of the bec in an optical periodic potential , and in particular that for the lowest band containing its phase excitations , can be accurately approximated by harmonic - oscillator wave functions .
these functions are simply described by the curvature of the on - site well as renormalized by the interactions in the bec and allow an explicit , though approximate , connection to be made between band modes and single - well modes .
they also allow very simple calculations on a periodic bec , as illustrated in section 3 for the dispersion relation in the lowest energy band and for the average wavenumber - dependent momentum in correspondence to a particular set of system parameters which is relevant to experiment . *
acknowledgments*. this work is supported by the istituto nazionale di fisica della materia through the advanced research project on bec .
we thank professor g. iadonisi for bringing slater s work to our attention .
the equation obeyed by the wave function @xmath123 at @xmath54 , that is @xmath124 , is @xmath125 we have chosen @xmath126 relative to the chemical potential @xmath25 of the condensate ( see the definition of @xmath127 immediately below eq .
( [ gpe ] ) ) . in view of the macroscopic occupation of this quantum state
, one must allow for fluctuations in the number of particles and for the accompanying shifts of the phase of the condensate .
this is accounted for by adding in the hamiltonian a `` kinetic energy '' term @xmath128 , where @xmath129 $ ] is the number operator generating such a `` translation '' and @xmath130 plays the role of a mass .
the canonically conjugate operator @xmath131 $ ] is easily seen to be the phase operator .
it satisfies the appropriate commutation relation provided that the weighting function @xmath132 obeys the equation @xmath133 with the condition @xmath134 . from a more formal viewpoint [ 19 , 20 ] ,
the bogolubov equations ( [ bogoeq ] ) at zero energy admit the eigenvector @xmath135 .
a second solution is provided by the eigenvector of the squared bogolubov operator with zero eigenvalue .
this implies that the result of applying the bogolubov operator to the second solution is proportional to @xmath135 through a constant to be determined , leading to eq .
( [ z00- ] ) .
in fact , upon differentiation of the gross - pitaevskii equation with respect to the total number of particles @xmath39 one is led again to eq .
( [ z00- ] ) , with @xmath136 and @xmath137 .
clearly , the constant @xmath32 is related to the change induced in the chemical potential @xmath25 by a fluctuation in the number of particles .
for the 1d mathieu problem ( non - interacting particles moving in the potential @xmath138 , slater [ 14 ] shows that the eigenfunctions @xmath139 in the momentum representation obey the adimensional finite - difference equation @xmath140=0\ ; , \ ] ] where @xmath141 , @xmath142 and @xmath143 . in the low - lying energy bands
@xmath144 is a slowly varying function of its argument , so that eq .
( [ slater1 ] ) is reduced to the differential equation @xmath145 evidently , this is the schrdinger equation for a harmonic oscillator , with energy levels @xmath146 and eigenfunctions @xmath147 where @xmath148 and @xmath149 are the hermite polynomials .
the wannier functions appropriate to the low - lying energy bands of the 1d mathieu problem are , therefore , single - well orbitals with the same functional form as for the harmonic oscillator . from his study of the exact wannier functions of the same problem slater [ 14 ] then showed that these single - well orbitals provide an accurate decription of the first five bands under the restriction @xmath150
i.e. @xmath79 .
the condition of validity for the first wannier function in eq .
( [ gauss ] ) can be estimated from slater s numerical results as @xmath80 . to extend this argument to the wannier functions of an interacting bec
, we view the mean - field interactions as equivalent to a renormalization in the height of the external potential , by an amount that we may estimate as the average value of @xmath151 over a lattice cell . from the normalization condition
satisfied by @xmath28 , in the case of repulsive interactions the height of the periodic potential is then _ lowered _ by an amount @xmath152 .
this leads to eq .
( [ slaterineq ] ) in the main text .
the operators @xmath153 entering the bogolubov equations ( [ bogoeqza ] ) and ( [ bogoeqzb ] ) can be split into an on - site part and a residual part as @xmath154 , where @xmath155 @xmath156 and @xmath157 @xmath158 in eqs .
( [ lsmeno]-[deltalpiu ] ) @xmath159 is the single - well potential @xmath160 , @xmath161 being the step function , @xmath162 is the on - site mean - field potential @xmath163 and @xmath91 is the on - site chemical potential , while the @xmath164 s are differences between lattice and single - well values .
namely , @xmath165 and @xmath166 a crucial quantity in the calculations reported in section 3.2 is the width @xmath102 of @xmath105 , since the integrals entering eq .
( [ bandlow ] ) are determined by the overlap in adjacent wells .
a first estimate of @xmath102 can be obtained from the curvature of the bottom of the well : this determines a frequency @xmath167 such that @xmath168 .
a more refined variational estimate can be based on a method first proposed by baym and pethick [ 16 ] , including the effect of interactions in the condensate .
we construct the energy functional @xmath169={\hbar^2\over 2m}\int dz\ ; \left|{d\phi(z)\over dz}\right|^2 + \int dz\ ; w(z)|\phi(z)|^2+\frac{1}{2}g\int dz\ ; |\phi(z)|^4\ ] ] and require that it be minimized by @xmath105 when we set @xmath170 .
this yields the energy function @xmath171 which is minimized when @xmath102 obeys the condition @xmath172 evidently , repulsive interactions broaden the wannier function .
a similar estimate of @xmath102 for the @xmath98 band states shows that the role of the interactions is decreasing with increasing energy .
q. niu , x .-
zhao , g. a. georgakis and m. g. raizen , phys .
76 * , 4504 ( 1996 ) .
s. r. wilkinson , c. f. bharoucha , k. w. madison , q. niu and m. g. raizen , phys .
lett . * 76 * , 4512 ( 1996 ) .
m. ben dahan , e. peik , j. reichel , y. castin and c. salomon , phys .
lett . * 76 * , 4508 ( 1996 ) .
e. peik , m. ben dahan , i. bouchoule , y. castin and c. salomon , phys .
a * 55 * , 2989 ( 1997 ) . c. f. bharoucha , k. w. madison , p. r. morrow , s. r. wilkinson , b. sundaram and m. g. raizen , phys .
a * 55 * , r857 ( 1997 ) .
b. p. anderson and m. a. kasevich , science * 281 * , 1686 ( 1998 ) . k.
berg - srensen and k. mlmer , phys .
a * 58 * , 1480 ( 1998 ) .
d. jaksch , c. bruder , j. i. cirac , c. w. gardiner and p. zoller , phys .
* 81 * , 3108 ( 1998 ) .
marzlin and w. zhang , phys .
a * 59 * , 2982 ( 1999 ) . m. j. steel and w. zhang , cond - mat/9810284 .
choi and q. niu , phys .
82 * , 2022 ( 1999 ) .
m. l. chiofalo , s. succi and m. p. tosi , phys . lett .
a * 260 * , 86 ( 1999 ) .
m. m. cerimele , m. l. chiofalo , f. pistella , s. succi and m. p. tosi , in the course of publication .
j. c. slater , phys . rev . * 87 * , 807 ( 1952 ) .
a. d. jackson , g. m. kavoulakis and c. j. pethick , phys . rev . a * 58 * , 2417 ( 1998 ) .
g. baym and c. j. pethick , phys .
lett . * 76 * , 6 ( 1996 ) . see e.g. a. griffin , phys .
b * 56 * , 9341 ( 1996 ) . see e.g. s. stringari , phys .
lett . * 77 * , 2360 ( 1996 ) .
j. p. blaizot and g. ripka , _ quantum theory of finite systems _ , mit press , cambridge ( 1986 ) .
m. lewenstein and l. you , phys .
lett . * 77 * , 3489 ( 1996 ) ; p. villain , m. lewenstein , r. dum , y. castin , l. you , a. imagoglu , t. a. b. kennedy , jou . of mod .
optics * 44 * , 1775 - 1799 ( 1997 ) .
see for instance j. m. ziman , _ principles of the theory of solids _ , cambridge university press , london ( 1972 ) .
p. r. wallace , phys . rev . * 71 * , 622 ( 1947 ) . | we study the dispersion relation of the excitations of a dilute bose - einstein condensate confined in a periodic optical potential and its bloch oscillations in an accelerated frame .
the problem is reduced to one - dimensionality through a renormalization of the s - wave scattering length and the solution of the bogolubov - de gennes equations is formulated in terms of the appropriate wannier functions .
some exact properties of a periodic one - dimensional condensate are easily demonstrated : ( i ) the lowest band at positive energy refers to phase modulations of the condensate and has a linear dispersion relation near the brillouin zone centre ; ( ii ) the higher bands arise from the superposition of localized excitations with definite phase relationships ; and ( iii ) the wavenumber - dependent current under a constant force in the semiclassical transport regime vanishes at the zone boundaries .
early results by j. c. slater [ phys . rev . * 87 * , 807 ( 1952 ) ] on a soluble problem in electron energy bands are used to specify the conditions under which the wannier functions may be approximated by on - site tight - binding orbitals of harmonic - oscillator form . in this approximation
the connections between the low - lying excitations in a lattice and those in a harmonic well are easily visualized .
analytic results are obtained in the tight - binding scheme and are illustrated with simple numerical calculations for the dispersion relation and semiclassical transport in the lowest energy band , at values of the system parameters which are relevant to experiment .
keywords : phase coherent atomic ensemble ( bose condensation ) ; quantum statistical theory : ground state and elementary excitations ; bloch oscillations . @xmath0 corresponding author .
tel . : + 39 - 050 - 509275 ; fax : + 39 50 563513 ; e - mail : [email protected] . |
the light curves of gamma - ray bursts ( grbs ) typically have many random peaks and in spite of extensive statistical studies ( e.g. , nemiroff et al .
1994 ; norris et al . 1996 ; stern 1996 ) , the temporal behavior of grbs remains a puzzle .
contrary to the complicated diverse behavior in the time domain , long grbs show a simple behavior in the fourier domain ( beloborodov , stern , & svensson 1998 ) .
their pds is a power law of index @xmath6 ( with a break at @xmath2 hz ) plus standard ( exponentially distributed ) statistical fluctuations superimposed onto the power law .
the pds slope and the break characterize the process randomly generating the diverse light curves of grbs .
intriguingly , the pds slope coincides with the slope of the kolmogorov law .
the power - law behavior is seen in individual bursts ( we illustrate this in 3 ) and may provide a clue to the nature of grbs . in the present paper
, we study in detail the pdss of gamma - ray bursts . in our analysis
we use 527 grb light curves with 64 ms resolution obtained by the burst and transient source experiment ( batse ) in the four lad energy channels , i iv : ( i ) @xmath7 kev , ( ii ) @xmath8 kev , ( iii ) @xmath9 kev , and ( iv ) @xmath3 kev .
the background is subtracted in each channel using linear fits to the 1024 ms data .
the method of data analysis is described in 2 . in 3 , we study a sample of the 4 brightest and longest bursts . in
4 the average pds for the full sample of 527 grbs is calculated and discussed . in 5 , we calculate the pdss in the separate energy channels and quantify the difference of the temporal structure between the channels in terms of the pds slope .
we also calculate the autocorrelation function ( acf ) in the separate channels and compare the results with previous studies of the acf . in 6 , we address dim grbs and compare them to the bright ones .
the results are discussed in 7 .
long bursts are of particular interest since their internal temporal structure can be studied by spectral analysis over a larger range of time scales .
we choose bursts with durations @xmath10 s where @xmath11 is the time it takes to accumulate from 5% to 95% of the total fluence of a burst summed over all the four channels , i+ii+iii+iv .
hereafter , we measure the brightness of a burst by its peak count rate , @xmath12 , in channels ii+iii . we analyze bursts with @xmath13 counts per time - bin , @xmath14 s. this condition coupled with the duration condition , @xmath10 s , gives a sample of 559 grbs .
the sample still contains grbs with low fluence , which are not good for fourier analysis .
we exclude bursts with fluences @xmath15 .
the resulting sample contains 527 bursts .
we calculate the fourier transform , @xmath16 , of each light curve , @xmath17 , using the standard fast fourier transform method .
the power density spectrum , @xmath18 , is given by @xmath19 since @xmath17 is real .
the fourier transform is calculated on a standard grid with a time bin @xmath20 ms and a total number of bins @xmath21 , which corresponds to a total time @xmath22 s since the trigger time .
the light curve of each burst is considered in its individual time window @xmath23 ( see 2.4 ) . in the calculations of the pds
, we extend the time interval to @xmath24 by adding zeros at @xmath25 and @xmath26 . the adding zeros introduces random small - scale fluctuations in the pds ( see , e.g. , bracewell 1967 ) .
we , however , study the pdss averaged over adjacent frequencies and/or over a sample of grbs .
then the fluctuations in @xmath18 associated with a specific choice of the grid disappear and do not affect the results .
poisson noise in the measured count rate affects the light curve on short time scales , and , correspondingly , affects the pds at high frequencies .
the poisson noise has a flat spectrum , @xmath27 , introducing the `` poisson level '' , @xmath28 , in a pds .
this level equals the burst total fluence including the background in the considered time window .
the power spectrum above this level displays the intrinsic variability of the signal ( but see 4.2 for the time - window effects ) .
we calculate the individual poisson level for each burst and subtract it from the burst pds . one would like to see the whole burst in the window .
however , ( 1 ) it is difficult to determine exactly the end of a burst because the burst can always have a weak `` tail '' hidden in the background ; ( 2 ) the window should not be very large because inclusion of long weak tails leads to an increase of the background fluence , @xmath29 , without a substantial increase in the signal fluence , @xmath30 .
the resulting low ratio @xmath31 implies a high poisson level , which makes the quality of the pds worse . to reduce the poisson level , we cut off the light curves at a time @xmath32 defined so that the signal count rate @xmath17 does not exceed @xmath33 at @xmath34 and @xmath35 . keeping in mind bursts with a weak beginning ,
we define the starting window time , @xmath36 , so that @xmath37 at @xmath38 and @xmath39 . in our analysis , @xmath40 is chosen .
we checked that varying @xmath41 does not significantly affect the results unless @xmath42 .
note that , for dim bursts , poisson fluctuations of the background ( which are imprinted on @xmath17 even after subtraction of the average background level ) may exceed @xmath43
. then the window @xmath23 is determined by the background fluctuations rather than by the signal .
the brightest and longest bursts are the best ones for fourier analysis . in this section ,
we study the four brightest bursts with @xmath0 s. they have trigger numbers # 2156 ( grb 930201 ) , 2856 ( grb 940302 ) , 3227 ( grb 941008 ) , and 6472 ( grb 941110 ) .
we take the light curves , @xmath17 , summed over channels ii and iii , in which the signal is strongest . to simplify the comparison of different bursts , we take peak - normalized light curves .
their fourier transform , @xmath16 , is therefore normalized by @xmath12 , and the pds , @xmath44 , is normalized by @xmath45 .
the four peak - normalized light curves and their pdss are shown in figure 1 ( we smooth the pdss on the scale @xmath46 before plotting ) .
the light curves are very different while the pdss are similar .
they can be described as a single power law , @xmath47 ( @xmath48 and @xmath49 ) with super - imposed fluctuations @xmath50 . in spite of the large @xmath51
, the power - law behavior can be seen in each burst due to the power - law extending over more than two decades in frequency .
the presence of an underlying power law is an interesting feature of the grb temporal behavior .
a possible way to extract it from the noisy individual pdss is to take the average pds over a sample of long grbs .
then the fluctuations affecting each individual pds tend to cancel each other and one can see the power - law behavior .
the average of the four pdss is plotted in figure 2 .
one can see that the amplitude of irregularities is reduced after the averaging and one can measure the slope of the resulting pds .
the power - law fit in the range @xmath52 gives @xmath53 .
( we use the standard @xmath54 fitting procedure with equal weights for each frequency bin . )
= 8.5cm=8.cm
the pds analysis was previously performed for a number of individual bursts .
belli ( 1992 ) , for example , analyzed grbs detected by the konus experiment onboard the soviet probes venera 11 and venera 12 , and giblin et al .
( 1998 ) studied individual batse bursts .
the power - law pds can be observed in their longest complex bursts as well .
when turning to grbs with relatively modest durations , @xmath55 s , it becomes more difficult to see the power - law behavior in individual bursts .
the statistical fluctuations , @xmath50 , significantly affect the apparent pds slope derived for the available range of frequencies , which is limited by @xmath56 .
however , having a large number of bursts allows one to study the pds by averaging over the sample .
in most of the grb models ( e.g. , in the internal shock model ) , different bursts are produced by one physical mechanism of stochastic nature , i.e. , an individual burst is a random realization of the same standard process .
stern & svensson ( 1996 ) suggested an empirical pulse - avalanche model of this type .
they showed that near - critical avalanches reproduce the statistical characteristics of grbs as well as their extreme diversity .
with such an approach , individual grbs are like pieces of a `` puzzle '' , and the features of the standard engine can be probed with statistical methods applied to a sufficiently large ensemble of grbs .
the simplest statistical characteristic of the pds is the average pds , @xmath57 .
the averaging means that we sum up the pdss of individual bursts with some weights ( normalization ) and divide the result by the number of bursts in the sample .
if no normalization is employed , then the brightest bursts strongly dominate @xmath57 , and their individual fluctuations lead to strong fluctuations in @xmath57 .
the pdss of relatively weak bursts are suppressed proportionally to @xmath58 and they are lost in the averaging .
a normalization procedure is thus needed to increase the weight of relatively weak bursts in the sample .
we prefer the peak - normalization because : \(i ) the amplitude of the resulting pds is practically independent of the burst brightness ( see , e.g. , fig . 1 ) .
\(ii ) the fluctuations in @xmath57 turn out to be minimal and the best accuracy of the slope is achieved .
the distribution of the individual @xmath18 around @xmath57 follows a standard exponential law ( see beloborodov et al .
1998 ) , and the amplitude of fluctuations in @xmath57 is given by a simple formula , @xmath59 , where @xmath60 is the number of bursts in the sample .
the averaging procedure is the following .
( 1 ) for each burst we first determine its individual poisson level .
( 2 ) we determine the peak , @xmath12 , of the light curve , @xmath17 , with the background subtracted .
( 3 ) we normalize the light curve to its peak . ( 4 ) we calculate the pds of the normalized light curve . ( 5 ) we normalize the poisson level by @xmath45 and subtract it from the pds . ( 6 ) we sum up the resulting pdss over the sample and divide by the number of bursts .
the resulting average pds , @xmath57 , for the 527 peak - normalized light curves in channels ( ii+iii ) is shown in figure 3 . for comparison
, we also plot the average pds with the @xmath61-normalization , where @xmath30 is the burst fluence excluding the background .
each light curve is then normalized by @xmath62 , and its pds ( and its poisson level ) is normalized by @xmath30 .
this normalization gives different weights ( the weights of dim bursts are reduced as compared to the peak - normalization ) .
the resulting average pds has a different amplitude .
nevertheless , @xmath57 again follows a power law with approximately the same slope .
this provides evidence that the self - similar behavior with @xmath6 is an intrinsic property of grbs , rather than an artifact of the averaging procedure .
this interpretation is also supported by the fact that we observe the power law in the longest individual bursts .
when comparing individual pdss , @xmath18 , with the peak - normalized @xmath57 , one sees that @xmath18 is exponentially distributed around @xmath57 , and the distribution is self - similar with respect to shifts in @xmath63 .
the power - law fit to @xmath57 in the range @xmath64 gives @xmath65 for the peak - normalized bursts , and @xmath66 for @xmath61-normalized bursts .
the formal error found from the @xmath54 fitting is very small , @xmath67 ( we take the variance , @xmath68 ) .
note that the slope in the peak - normalized case is different from @xmath69 reported in beloborodov et al .
this change is caused by the fact that we have extended the sample to smaller brightnesses ( see 6 ) .
the deviation from the power law at the low - frequency end is due to the finite duration of bursts . at the high frequency end
, there is a break at @xmath2 hz .
the break is observed in the brightest grbs even without subtracting the poisson level ( see beloborodov et al .
1998 and the top panel in fig . 8) .
note that the break position is the same for the peak - normalized and @xmath61-normalized bursts ( see fig .
it stays the same in the separate energy channels ( see fig
. 5 ) and does not depend on the poisson level .
one may also see the break in individual long bursts ( figs . 1 and 2 ) .
the break is far too sharp to be explained as an artifact of the 64 ms time binning , which suppresses the pds by a factor of @xmath70 ^ 2 $ ] where @xmath20 ms is the time bin ( cf .
van der klis 1989 ) .
= 8.0 cm = 8.0 cm fourier analysis was designed for physical problems dealing with linear differential equations .
for example , it is usually applied to small perturbations above a given background solution .
the fourier power spectrum is also commonly used in the temporal studies of long signals or noises , e.g. , in persistent astrophysical sources . by contrast
, grbs have strongly non - linear signals with short durations .
the typical number of batse time - bins ( @xmath20 ms ) in a long grb is a few @xmath71 .
do the effects of finite duration ( i.e. , time - window effects ) strongly affect the measured pds ? the issue is illustrated in figure 4 .
we prepared an artificial long signal with @xmath72 and exponentially distributed @xmath18 , i.e. , the probability to detect @xmath18 at a given @xmath63 is proportional to @xmath73 ) .
the phase structure was taken to be gaussian ( i.e. , random ) .
the signal duration is @xmath74 s. then , we cut the long signal into 64 pieces of equal length @xmath75 s. we thus get 64 short signals , each is a random realization / fragment of the same stationary process characterized by the index @xmath76 .
analogously to our analysis of real grbs , we normalize each signal to its peak and add zeros up to @xmath77 ( our standard grid ) .
then we calculate the average pds , @xmath57 , for the 64 artificial signals .
the result is shown in figure 4 for the three cases : @xmath78 ( poisson ) , @xmath79 ( flicker ) , and @xmath80 ( kolmogorov ) .
one can see that the time - window effects are strong in the case @xmath78 .
the poisson signal is roughly constant on large time scales . as a result , at modest @xmath63
, @xmath57 is just the power spectrum of the @xmath81-window .
by contrast , in the case @xmath80 , we have large - amplitude variations in the signal on all time scales .
the average pds of the 64 short signals then reproduces well the intrinsic pds slope throughout the whole range of frequencies , down to @xmath82 ( see bottom panel in fig . 4 ) .
hence , the power law we observe in the average pds of grbs can be interpreted as that grbs are random short realizations of a standard process which is characterized by the pds slope @xmath6 .
note , however , that the power spectrum does not provide a complete description of the signal since the phase structure is not considered .
what does the average pds look like in the separate energy channels ?
the signal to noise ratio is low in channel i and especially in channel iv .
the number of grbs for which good pdss can be obtained in channels i iv is therefore limited to the brightest bursts .
we choose a sample of bursts with @xmath83 counts / bin ( in channels ii+iii ) , which contains 152 bursts .
the burst analysis is now performed in each channel separately .
we determine the poisson level of a burst in each channel , @xmath84 , and find the peak of the light curve , @xmath85 , where @xmath86 i , ii , iii , iv . then we perform the peak - normalization : @xmath87 and @xmath88 .
we set the time window in each channel @xmath89 as described in 2.4 .
= 16.5cm=16.5 cm the resulting average pdss in channels i iv are shown in figure 5 .
the differences in the slopes are clearly seen .
we fitted the pdss by power laws , @xmath90 , in the range @xmath91 .
channel iv is fitted in the range @xmath92 .
@xmath93 and @xmath76 of the fits are listed in table 1 .
[ cols="^,^,^,^,^ " , ] in the pds calculations we subtracted the poisson level , which is quite high for dim bursts ( see fig .
one therefore should address a technical question : how well is the intrinsic pds restored after subtracting the poisson level ? to investigate this issue we have performed the following test .
we take the sample of 91 bright bursts ( group a ) and add a strong poisson noise ( with an average level of 25000 counts / bin ) to each burst in the group .
we thus artificially increase the poisson level by two orders of magnitude .
alternatively , one may consider this procedure as a rescaling of bright bursts to a smaller brightness while keeping the poisson background at the same level .
then , we analyze the artificial bursts in the same way as we did with the real dim grbs .
note that even after subtraction of the time - averaged background , its poisson noise strongly affects the signal and may create artificial peaks dominating the true peak of the signal .
for the artificial bursts , we know the true peak ( which is the peak of the original bright burst ) . in real dim bursts we do not know the position of the true peak and employ the peak search scheme described in stern , poutanen , & svensson ( 1999 ) .
= 14.5cm=14.5 cm the result is compared with the average pds of the original sample in figure 9 .
we find that the subtraction of the poisson level allows one to restore the original pds , @xmath18 , also at frequencies where @xmath18 is well below the poisson level .
even the 1 hz break remains present .
we conclude that the high poisson level is unlikely to significantly affect the measured pds slope .
the average time profile ( atp ) of grbs was found to follow a stretched exponential of index @xmath94 ( stern 1996 ) .
is there any relation between the atp and the average pds ?
one should note two important differences between the atp and the pds studies : ( 1 ) the @xmath5 pds , though affected by the statistical fluctuations , is observed in _
individual _ bursts , while the atp is a purely statistical property of a large sample of bursts ; ( 2 ) the atp is constructed for grbs of _ all durations _ ( and only in this case it displays the perfect stretched exponential ) while the pds is studied for long bursts only .
nevertheless , both the atp and the pds characterize the stochastic process generating grbs .
stern ( 1999 ) shows that both can be reproduced simultaneously with the pulse - avalanche model of stern & svensson ( 1996 ) .
the sharpness of the break in the average pds appears to be an important feature that constrains models of grbs .
if the signal is produced in the rest frame of a relativistic outflow then variations of the outflow lorentz factor , @xmath95 , from burst to burst would smear out the break .
the sharpness of the break then implies a narrow dispersion of @xmath95 , @xmath96 , which appears to be unlikely . finally , one may associate the break with the dynamical time scale corresponding to the inner radius of the optically thin zone of the outflow , @xmath98 .
then the variability on time scales shorter than @xmath99 can be suppressed . in this case , however , one should explain why @xmath100 stays the same in different bursts .
it has often been hypothesized that dim bursts are at high cosmological redshifts .
for instance , it is necessarily the case if grbs have approximately the same intrinsic luminosity , the so - called `` standard candle '' hypothesis .
we can test this hypothesis using the power spectrum analysis .
suppose that the dim bursts are intrinsically the same as the bright ones .
then any difference in their average pds should be due to a cosmological redshift .
first consider the bolometric light curves assuming that their average pds follows the @xmath5 power law .
it is easy to see that a redshift , @xmath101 , will not change the pds slope . as we normalize each bursts to its peak before the averaging , the effect of a redshift is just stretching the light curve in time , i.e. , precisely the time dilation effect
this will lead to an increase of the net normalization of the pds by a factor of @xmath102 .
the slope does not change since the dilation factor @xmath103 is the same for each time scale .
one should , however , recall that we observe bursts in a limited spectral interval .
a redshift then implies a shift of the signal with respect to our spectral window .
e.g. , photons detected in channel iii would originally have been emitted in channel iv .
as we know from 5 , the pdss are different in different energy channels .
therefore , one expects that the pds slope will change for redshifted grbs .
we have seen in 5 that the pds of bright bursts flattens in the hard channels .
hence , a redshift of the bright bursts must be accompanied by a flattening of the pds .
contrary to this behavior , we observe that the pds of dim bursts _
steepens_. hence , the evolution of the pds with brightness is inconsistent with the standard candle hypothesis .
it implies that the burst luminosity function is broad and dim bursts are intrinsically weak .
evidence for a broad luminosity function is also found when looking at the isotropic luminosities of the bursts with measured redshifts .
note , however , that the differences in the apparent luminosities could be caused by orientation effects if grbs are beamed
. one should not therefore exclude that the total intrinsic luminosities are approximately the same .
the different temporal structure of dim bursts may be an important fact in this respect .
in particular , it suggests that the observed time dilation of dim bursts ( e.g. , norris et al .
1994 ) may be caused mainly by physical processes occurring in the bursts rather than by a cosmological redshift .
note that the intrinsic difference of the temporal structure of dim grbs was also found when studying their average time profile ( see stern et at .
the rising part of the atp does not change with decreasing brightness while the decaying part suffers from time dilation .
this behavior is inconsistent with a cosmological time dilation which should apply equally to both parts of the atp .
a likely scenario of grbs involves internal shocks in a relativistic outflow with a lorentz factor @xmath104 ( see , e.g. , piran 1999 for a review ) .
the shock develops when an inner faster shell of the outflow tries to overtake the previous slower shell .
the pulses in a burst are then associated with collisions between the shells .
the pdss predicted by the model were recently tested against the observed @xmath5 law ( panaitescu , spada , & mszros 1999 ; spada , panaitescu , & mszros 2000 ; beloborodov 1999,2000 ) .
the self - similar temporal structure with @xmath80 can be reproduced by the model .
further constraints should , however , be imposed by the observed dependence of @xmath76 on photon energy . besides , the phase structure of the signal , neglected so far , should be taken into consideration ( beloborodov 2000 ) .
this research has made use of data obtained through the heasarc online service provided by nasa / gsfc .
we thank a.f .
illarionov , i. panchenko , and j. poutanen for discussions .
this work was supported by the swedish natural science research council , the swedish royal academy of science , the wennergren foundation for scientific research , a nordita nordic project grant , the swedish institute , rfbr grant 97 - 02 - 16975 , and nsf grant phy94 - 07194 . | power density spectra ( pdss ) of long gamma - ray bursts ( grbs ) provide useful information on grbs , indicating their self - similar temporal structure .
the best power - law pdss are displayed by the longest bursts ( @xmath0 s ) in which the range of self - similar time scales covers more than 2 decades .
shorter bursts have apparent pds slopes more strongly affected by statistical fluctuations .
the underlying power law can then be reproduced with high accuracy by averaging the pdss for a large sample of bursts .
this power law has a slope @xmath1 and a sharp break at @xmath2 hz .
the power - law pds provides a new sensitive tool for studies of grbs .
in particular , we calculate the pdss of bright bursts in separate energy channels .
the pds flattens in the hard channel ( @xmath3 kev ) and steepens in the soft channel ( @xmath4 kev ) , while the pds of bolometric light curves approximately follows the @xmath5 law .
we then study dim bursts and compare them to the bright ones .
we find a strong correlation between the burst brightness and the pds slope .
this correlation shows that the bursts are far from being standard candles and dim bursts should be _ intrinsically _ weak .
the time dilation of dim bursts is probably related to physical processes occurring in the burst rather than to a cosmological redshift . |
in a free quantum theory particle numbers of various species are conserved . however ,
if there is a time varying background or a source term in the system , particle number conservation gets broken and particles are produced by non - perturbative effects .
it is well known that an oscillating electric field causes production of pairs of electrons @xcite , and time - varying or inhomogeneous metric causes gravitational particle production @xcite . in this context
the period of post - inflationary preheating , when particles coupled to the inflaton field are produced on the expense of coherently oscillating inflaton background , is quite interesting for both particle physics and cosmology because the phenomenon not only marks the end of inflation but also produces matter constituting our observable universe @xcite . in this paper
we are interested in non - gravitational particle production with a signficant role of interaction ( non - quadratic ) terms .
the basic mechanism of particle production can be explained with the help of a simple model @xcite : @xmath0 where @xmath1 and @xmath2 are respectively a complex and a real bosonic field , and @xmath3 is a real coupling constant .
lagrangian ( [ eom ] ) may for example describe the coupling between inflaton @xmath1 and a standard model field @xmath2 resulting in production of visible particles . in this model @xmath4 becomes a special kind of point in the phase space where the other field becomes effectively massless .
has the same properties . ]
especially , in case of @xmath5 , particle vacuum can be realized for an arbitrary value of @xmath1 .
let us consider the case with @xmath1 having a time - dependent vacuum expectation value ( vev ) @xmath6 .
asymptotically , when quantum effects can be neglected , one can choose a vacuum solution of ( [ eom ] ) in the form of uniform motion : @xmath7 where a real @xmath8 represents a velocity in the @xmath1-space and a real @xmath9 is an impact parameter measuring the distance of the classical trajectory from the ( massless ) point @xmath10 .
when this trajectory approaches the massless point within the circle @xmath11 , which is the non - adiabaticity condition for @xmath2 , mass of @xmath2 becomes very small . as a result ,
kinetic energy of the background field @xmath1 is transferred to the field @xmath2 and this leads to the production of @xmath2 particles whose distribution is calculated to be @xcite,@xcite @xmath12 where @xmath13 is a momentum of a @xmath2 particle .
furthermore , an interesting phenomenon occurs after the initial period of @xmath2 particle production .
once particles @xmath2 are created , and trajectory of @xmath6 goes away from the massless point , energy density of @xmath2 particles can be represented in the lagrangian by a term which has a meaning of mass multiplied by number density , i.e. @xmath14 this implies that a linear potential is established in the field space .
due to the force exerted by this term in the lagrangian , after a while trajectory of @xmath6 may go back towards the origin and @xmath2 particles can be produced again . finally , after a number of cycles , trajectory @xmath6 may get trapped near the massless point .
our main goal in this paper is to extend above considerations to the case where interactions between fields are significant and may become a leading effect , for instance in the absence of the effective mass terms . as an illustrative example we take a simple supersymmetric generalisation of the model ( [ eom ] ) with unbroken global supersymmetry .
in such a case , fermionic particles @xmath15 and @xmath16 , which are superpartners of @xmath1 and @xmath2 , are added to the previous theory . as we will discuss later ,
if @xmath1 has a vev then @xmath2 and @xmath16 obtain masses proportional to @xmath17 , while @xmath15 stays massless .
it is easy to see that @xmath2 and @xmath16 particles are produced . on the other hand
, the mass of @xmath15 is not varying with time since it is absent from the lagrangian .
however , since the theory we consider has more interactions , e.g. @xmath18 , we shall need to understand their effect on particle production . as a result
, we will find that massless particle can in fact be produced due to interactions .
outline of the paper is as follows .
section 2 presents the general framework used to calculate number density of produced particles in case of existence of the interaction terms and quantum corrections in the theory ; general form of bogoliubov transformation is also described there .
section 3 focuses on application of the method described in section 2 to the particular supersymmetric model in both analytical and numerical approach .
rather technical subsection 3.1 defines precisely asymptotic fields and bogoliubov transformations in our model .
before we analyse a specific model , let us formulate the general recipe for calculation of the number of produced particles in the presence of quantum corrections .
at first we define an occupation number of particles with momentum @xmath13 as @xmath19 where @xmath20 is an in - state vacuum , and @xmath21 is an annihilation operator defined by an out - state of vaccum . in order to calculate this quantity ,
one employs usually the method of bogoliubov transformation . with this method one establishes the relation between @xmath22 and @xmath21 in the form @xmath23 where @xmath24 and @xmath25 are complex functions of @xmath26 .
however , calculation of the bogoliubov coefficients becomes tricky when theory has large interaction terms .
then , because the out - state is affected by other fields via interaction terms , the @xmath21 shall contain information about all field operators in the interacting model , not only about its own field operator . in this section
we will discuss how the bogoliubov coefficients are calculated in a complete interacting model .
let us begin with a real scalar field @xmath27 .
we assume that the scalar field operator satisfies the following commutation relation @xmath28 = i\delta^3(\mathbf{x}-\mathbf{y } ) \label{eq : psi_com_rel1}\ ] ] and equation of motion is of the form @xmath29 where @xmath30 is a mass of @xmath27 which is a classical number ( i.e. not an operator ) , and @xmath31 is a source term which is an operator consisting of @xmath27 or other fields . in general @xmath30 may be a function of space - time coordinates @xmath32 , but here we assume only time - dependence for simplicity : @xmath33 .
equation ( [ eq : psi ] ) has a formal solution known as _ yang - feldman equation _
@xcite given as @xmath34 j(y ) , \label{eq : psi_yf}\ ] ] where @xmath35 is called an asymptotic field which satisfies free field equation @xmath36 and @xmath37 denotes the field renormalization constant .
+ in general , source term @xmath31 is a function of interacting field operator .
however , one can obtain a sequentially - iterated result from this formal solution substituting ( [ eq : psi_yf ] ) back to @xmath31 again .
in particular , if we take @xmath38 , ( [ eq : psi_yf ] ) becomes @xmath39 therefore @xmath40 denotes a specific point of time when an interacting field @xmath27 can be regarded as a free field @xmath35 . using this relation
one can represent the out - field ( defined at @xmath41 ) by the in - field ( defined at @xmath42 ) as follows : @xmath43 j(y ) .
\label{eq : psi_in - out - rel}\ ] ] one can see that out - field operator depends not only on the in - field but also on the source term which contains interaction effects . since ( [ eq : psi_as_eom ] ) is a free field equation of motion , solution of @xmath35
can be expanded into wave functions as @xmath44 , \label{eq : psi_as_expand}\ ] ] where @xmath45 is an annihilation operator which , following equation ( [ eq : psi_as_eom ] ) , satisfies commutation relation @xmath46 = ( 2\pi)^3 \delta^3(\mathbf{k - k ' } ) , \qquad ( { \rm other \ : relations } ) = 0 \label{eq : psi_com_rel2}\ ] ] and @xmath47 ( where @xmath48 ) is a time dependent wave function which satisfies @xmath49 using commutation relations ( [ eq : psi_com_rel1 ] ) and ( [ eq : psi_com_rel2 ] ) , one can obtain a time - independent inner product relation ) and then uses equation of motion . ] for the wave function : @xmath50 with this formula one can extract @xmath45 from @xmath35 as follows : @xmath51 it means that a component along the @xmath47-direction is projected out of @xmath35 by means of the inner product ( [ eq : inner_pro ] ) .
relation ( [ ann ] ) for out - states can be obtained using ( [ eq : psi_in - out - rel ] ) as @xmath52 j(y ) \nonumber \end{aligned}\ ] ] or @xmath53 where time - independent coefficients @xmath24 and @xmath25 are defined as @xmath54 from these relations one learns how much of @xmath55 or @xmath56 can be found in @xmath57 .
thus one can regard @xmath24 and @xmath25 as the bogoliubov coefficients , which satisfy the following properties : @xmath58 and proper normalization @xmath59 hence , formulae ( [ eq : bbt ] ) and ( [ eq : bt_a ] ) represent a bogoliubov transformation law generalized to the case of interacting theory .
first two terms in these expressions correspond to the usual bogoliubov transfomation law arising from the time ( or space ) variation of masses in the lagrangian and has nothing to do with the source operator @xmath60 .
the third term describes the effects coming from interaction terms ( quantum corrections ) .
finally , expression ( [ eq : bt_a ] ) enables us to establish the formula for the occupation number of produced particles in case of interacting theory @xmath61 where @xmath62 is the volume of the system .
this result shows that because of the interaction effects particle production can be induced in both cases , @xmath63 and @xmath64 , even though the leading term of @xmath65 for @xmath63 is equal to zero .
+ analogous results hold for fermions , with a different commutation relation .
in this section we consider production of particles in a specific model . for the sake of simplicity we assume unbroken supersymmetry and the superpotential @xmath66 where @xmath67 and @xmath68 are chiral superfields which interact with a coupling strength @xmath3 .
corresponding interaction lagrangian reads @xmath69 where @xmath70 and @xmath71 are the scalar / fermion components of @xmath67 and @xmath68 supermultiplets respectively .
the massless point is @xmath72 and @xmath1 can have an arbitrary value there .
therefore if @xmath1 has a vev @xmath73 , then @xmath2 and @xmath16 are given a mass @xmath74 but as we quantize a deviation from the vev , @xmath75 , @xmath76 and @xmath15 remain massless .
there are two simplifications when one restricts oneself to this toy model .
the first one is that we can calculate number of produced particles without worrying about mass counterterms due to supersymmetry .
the second is that this is a minimal supersymmetric model which contains not only fields with time - varying masses but also massless fields .
these facts indicate that the bogoliubov coefficient @xmath25 for massive fields ( @xmath77 ) becomes non - zero , on the other hand , @xmath63 for the massless fields ( @xmath78 ) .
as shown in the previous section , it is possible to produce these massless particles even if @xmath63 due to interaction effects .
therefore we can compare the produced particle number in all cases between @xmath64 and @xmath63 .
in this , rather technical , section we demonstrate derivation of the bogoliubov transformation laws for all component fields present in the lagrangian ( [ eq : lagrangian ] ) .
the procedure follows steps outlined in section [ sec : qe ] . from ( [ eq : lagrangian ] ) one obtains a set of usual equations of motion : @xmath79 with source terms given as @xmath80 in the above formulae we took into account the possible influence of the vev @xmath6 on @xmath2 and @xmath16 masses . following ( [ source ] )
we can obtain the yang - feldman equations for scalar fields as @xmath81 j_{\phi}^{\dagger}(y ) , \label{eq : formal_phi } \\
\chi(x ) & = & \sqrt{z_{\chi } } \chi^{\rm as}(x ) - i z_{\chi } \int_{t^{\rm
as}}^{x^0 } dy^0 \int d^3y \left [ \chi^{\rm as}(x ) , \chi^{{\rm as}*}(y ) \right ] j_{\chi}^{\dagger}(y).\end{aligned}\ ] ] analgous equations for fermions read @xmath82 asymptotic fields , by definition , can be regarded as free fields what enables us to expand them into plane waves .
such decomposition for bosons differs only by a vev ( @xmath83 while @xmath84 ) @xmath85 whereas for fermions we are obliged to take into account also helicity ( see appendix ) @xmath86 here superscript `` as '' denotes the wave functions satisfying the following set of equations of motion @xmath87 and one has commutation relation for scalars and anticommutation relation for fermions of the form @xmath88 = \left [ a_{\chi \mathbf{k}}^{s,{\rm as } } , a_{\chi \mathbf{k'}}^{r,{\rm as } \dagger } \right ] = ( 2\pi)^3 \delta^3(\mathbf{k - k ' } ) \delta^{sr},\ ] ] @xmath89 in the above formulae @xmath90 labels distinguish between a particle and antiparticle for scalars and indicate the helicity of fermions . using the established commutation relation
one can obtain inner product relations obeyed by the components of wave functions for all species : @xmath91 for scalars and @xmath92 for fermions . with these products
one can extract the expressions for annihilation operators using respective yang - feldman equations with @xmath93 : @xmath94 , \label{eq : a_phi } \\ a_{\phi \mathbf{-k}}^{-,{\rm out}\dagger } & = & + iz \int d^3x e^{-i\mathbf{k \cdot x } } \left [ \dot{\phi}_k^{\rm out } \left ( \phi^{\rm out } - \left < 0^{\rm out } \left| \phi^{\rm out } \right| 0^{\rm out } \right >
\right ) \right .
\nonumber \\ & & \qquad \qquad \qquad \qquad \left .
- \phi_k^{\rm out } \left ( \dot{\phi}^{\rm out } - \left < 0^{\rm out } \left| \dot{\phi}^{\rm out } \right| 0^{\rm out } \right >
\right ) \right ] , \nonumber \\
\mathbf{k}}^{+,{\rm out } } & = & -iz_{\chi } \int d^3x e^{-i\mathbf{k \cdot x } } \left ( \dot{\chi}_k^{{\rm out } * } \chi^{\rm out } - \chi_k^{{\rm out } * } \dot{\chi}^{\rm out } \right ) , \nonumber \\ a_{\chi \mathbf{-k}}^{-,{\rm out}\dagger } & = & + iz_{\chi } \int d^3x e^{-i\mathbf{k \cdot x } } \left ( \dot{\chi}_k^{\rm out } \chi^{\rm out } - \chi_k^{\rm out } \dot{\chi}^{\rm out } \right ) , \nonumber \\ a_{\psi_{\phi } \mathbf{k}}^{+,{\rm out } } & = & z_{\psi_{\phi } } \int d^3x e^{-i\mathbf{k \cdot x } } \psi_{\phi k}^{{\rm out } * } \cdot e_{\mathbf{k}}^{+\dagger } \bar{\sigma}^0 \psi_{\phi}^{\rm out } , \nonumber \\
a_{\psi_{\phi } \mathbf{-k}}^{-,{\rm out}\dagger } & = & z_{\psi_{\phi } } \int d^3x e^{-i\mathbf{k \cdot x } } \psi_{\phi k}^{\rm out } \cdot e_{\mathbf{k}}^{-\dagger } \bar{\sigma}^0 \psi_{\phi}^{\rm out } , \nonumber \\ a_{\psi_{\chi } \mathbf{k}}^{s,{\rm out } } & = & z_{\psi_{\phi } } \int d^3x e^{-i\mathbf{k \cdot x } } \left ( \psi_{\chi k}^{(+)s,{\rm out } * } \cdot e_{\mathbf{k}}^{s\dagger } \bar{\sigma}^0 \psi_{\chi}^{\rm out } \right .
\nonumber \\ & & \left .
\qquad \qquad \qquad \qquad \qquad + s e^{-i\theta_{\mathbf{k } } } \psi_{\chi k}^{(-)s,{\rm out } } \cdot \psi_{\chi}^{{\rm out}\dagger } \bar{\sigma}^0 e_{\mathbf{-k}}^s \right ) .
\label{eq : a_psi_chi } \nonumber\end{aligned}\ ] ] also the relation between in and out states , leading to the bogoliubov transformation , can be found : @xmath95 j_{\phi}^{\dagger}(y ) , \\
\chi^{\rm out}(t^{\rm out } , \mathbf{x } ) & = & \chi^{\rm in}(t^{\rm out } , \mathbf{x } ) - i \sqrt{z_{\chi } } \int d^4y \left [ \chi^{\rm in}(t^{\rm out } , \mathbf{x } ) , \chi^{{\rm in}\dagger}(y ) \right ] j_{\chi}^{\dagger}(y),\nonumber \end{aligned}\ ] ] for scalars and @xmath96 for fermions .
putting together available information we can now derive the following bogoliubov tranformation rules for scalar @xmath1 : @xmath97 and for scalar @xmath2 : @xmath98 in the fermionic case one finds for the @xmath15 field : @xmath99 and for the @xmath16 : @xmath100 the bogoliubov coefficients for massive fields are given by @xmath101 for scalars and @xmath102 for fermions .
it should be noted , that the usual bogoliubov coefficients are absent in case of massless field @xmath1 and @xmath15 .
the reason is that their mass terms do not change with time and their equations of motion can be solved over the whole time interval .
therefore the form of wave functions in the in - state and in the out - state is the same , i.e. @xmath103 and @xmath104 , hence the bogoliubov coefficients @xmath25 for @xmath1 and @xmath15 should vanish indeed . for massive particles @xmath2 and @xmath16 , the leading results for occupation numbers are calculated , using the framework described in section 2 , to be : @xmath105 where @xmath62 is the volume of the system . the factor 2 in ( [ eq : chi_occu ] ) counts degrees of freedom for a complex scalar field .
if one takes @xmath106 as the initial condition , one obtains @xmath107 using the analytical continuation method as in ( [ n ] ) .
the standard quasi - classical reasoning leads to the same result for bogoliubov coefficients in case of fermions and in case of bosons . as a consequence
, we can evaluate the number densities to be @xmath108 on the other hand , in the case of production of massless particles @xmath76 and @xmath15 interaction effect in the bogoliubov transformation formula is the leading contribution , since the respective bogoliubov coefficient @xmath109 is zero .
up to the lowest power of the coupling @xmath3 , one finds for the massless scalar : @xmath110 , \label{eq : phi_occu } \end{aligned}\ ] ] and for the massless fermion : @xmath111 the above formulae look complicated but we can find the physical meaning by considering perturbation theory diagrams . since the wave function corresponds to an external line of a diagram ,
the diagrams corresponding to ( [ eq : phi_occu ] ) and ( [ eq : psi_phi_occu ] ) are the ones shown in the figure [ fig : diagrams ] .
note that these diagrams can be described as `` inverse decay '' processes .
one might suspect that they are forbidden because of conservation of energy and momentum .
however , these processes are actually possible since the background field is time - dependent and one needs to take into account energy non - conservation in the diagrams representing quantum processes in this theory .
therefore , at least from the perturbative point of view , we can conclude that the main effect in the massless particle production comes from inverse decay processes of massive particles in varying external background .
as far as the loop corrections are concerned , their effect on particle production in case of the weak coupling is supressed by the growing powers of g multiplied by small numerical loop - factors . in case of a strong coupling
this effect becomes much more subtle - loop contributions to the propagators and vertices considered here can become large and even growing with time , as argued in @xcite and @xcite .
discussion of potentially large loop effects goes beyond the scope of this paper and shall be addressed elsewhere .
it is difficult to evaluate formulae ( [ eq : phi_occu ] ) and ( [ eq : psi_phi_occu ] ) analytically ( we will show numerical results in section [ sec : numerical_results ] ) .
however , we can find the behaviour of number densities for massless particles at low momentum ( or large vev ) if we use rough approximations @xmath112 this clearly corresponds to the exponential supression of ( [ eq : num_chi_psi_chi ] ) .
the above approximations simplify ( [ eq : phi_occu ] ) and ( [ eq : psi_phi_occu ] ) to the form @xmath113 we will use these results later to make fits to numerical results ) shows a quadratic time dependence , which is supported by our numerical results , there are papers which claim a linear time dependence , e.g. see @xcite . ] .
as we mentioned before analytical approach is severely limited in case of particle production in interacting theories .
equations to be solved are those determining @xmath114 and in - state wave functions for all available species . for massless ones we can find analytically that @xmath115 but for the massive case analytical solution is impossible and one is left with wkb type approximate solutions @xmath116 where @xmath117 .
wkb approximation is valid provided the condition @xmath118 is fulfilled . once one starts with a large enough vev @xmath6
, one can safely choose for these solutions initial conditions in the form : @xmath119 @xmath120 we can also assume wkb type solutions also for the out - state wave functions : @xmath121 it is justified because we are still far away from the non - adiabatic region in the phase space .
determining the vev of @xmath1 we consider the backreaction at the level of the asymptotic field expansion . limiting ourselves to the 1-loop level in momentum integration
we obtain the following equation fixing @xmath122 : @xmath123 putting together all pieces of information we can obtain approximate expressions for number densities of produced particles : @xmath124 equations describing number density of produced particles , ( [ eq : phi_occu ] ) , ( [ eq : psi_phi_occu ] ) , ( [ n_chi ] ) and ( [ n_psi_chi ] ) , can not be solved accurately analytically but are not so difficult to be solved numerically . numerical results for number densities are shown in the figure [ fig : number_density_g=1 ] .
( right vertical axis ) and number densities ( left vertical axis ) .
we took @xmath125 , @xmath126 , @xmath127 .
values of number density at @xmath128 are : @xmath129 , @xmath130 , @xmath131 , @xmath132 . ]
one can see that particle production occurrs not only for massive particles but also for the massless ones , as expected . for the produced number of massive species ,
numerical results show a good agreement with analytic ones .
for reasonable values of free parameters , @xmath133 , one finds analytically : @xmath134 , what is in a good agreement with our numerical results . comparing the massless particle number with massive one , for the above set of parameters , the ratios after the period of production , at @xmath128 ,
are @xmath135 for bosons and @xmath136 for fermions . for comparison , in figure [ fig : number_density_g=2 ]
we also show our numerical results for stronger coupling .
weak and strong couplings have the usual meaning in this paper but crucial difference between them is that stronger coupling leads to the trapping effect mentioned earlier while weak coupling does not . and number densities in case of stronger coupling ( @xmath137 ) .
other parameters are the same as in the weaker case in figure [ fig : number_density_g=1 ] . ] in the case of strong coupling , massless bosons @xmath1 are produced as abundantly as massive bosons @xmath2 .
one can also observe explicitly the second round of particle production which is caused by the trapping effect for @xmath6 .
one might wonder why the produced number density of @xmath1 becomes larger than that of massive particles .
this may be an artefact caused by limited accuracy of numerical calculations ( only up to terms of the order of @xmath138 ) .
nevertheless , it is safe to conclude that massless particles can be produced as abundantly as massive particles provided the necessary couplings are sizeable .
finally , we show numerical results of distribution function for massless species and their fitting functions as in figure [ fig : fittings ] .
the fitting functions are chosen as follows : @xmath139 ^ 2 , \label{eq : n_phi_k_fitting } \\
n_{\psi_{\phi } k } / v & \sim & 0.40 \cdot \frac{g^2}{4\pi } \frac{1}{e^{\sqrt{\pi k^2/g|v|}}+1 } \cdot \sqrt{g|v|}(t - t _ * ) \left [ \frac{\sin{0.59k(t - t_*)}}{0.59k(t - t _ * ) } \right]^2 , \label{eq : n_psi_phi_k_fitting}\end{aligned}\ ] ] where @xmath140 is the point of time when a trajectory of @xmath1 is closest to the massless point ( @xmath10 ) .
although these functions have been chosen in such a way that their behaviour at small momentum @xmath141 satisfies ( [ eq : n_phi_k_low_momemntum ] ) and ( [ eq : n_psi_phi_k_low_momemntum ] ) , they follow very faithfully the numerical results over the whole available range of momenta due to the additional sine function .
the structure of the fitting functions is quite interesting .
basically , they consist of the usual bosonic / fermionic distributions^{-1}$ ] drops at high momentum much faster than the numerical results . ] multiplied by perturbative suppression @xmath142 .
( thick lines ) and their corresponding fitting functions for various couplings @xmath143 , and @xmath144 ( thin lines ) .
the initial values for numerical results are @xmath125 and @xmath126 ( in figure [ fig : number_density_g=1 ] ) . for the fitting functions we set @xmath145 which are chosen to be consistent with initial conditions for numerical results.,title="fig : " ] ( thick lines ) and their corresponding fitting functions for various couplings @xmath143 , and @xmath144 ( thin lines ) .
the initial values for numerical results are @xmath125 and @xmath126 ( in figure [ fig : number_density_g=1 ] ) . for the fitting functions we set @xmath145 which are chosen to be consistent with initial conditions for numerical results.,title="fig : " ] ( thick lines ) and their corresponding fitting functions for various couplings @xmath143 , and @xmath144 ( thin lines ) .
the initial values for numerical results are @xmath125 and @xmath126 ( in figure [ fig : number_density_g=1 ] ) . for the fitting functions we set @xmath145 which are chosen to be consistent with initial conditions for numerical results.,title="fig : " ] ( thick lines ) and their corresponding fitting functions for various couplings @xmath143 , and @xmath144 ( thin lines ) .
the initial values for numerical results are @xmath125 and @xmath126 ( in figure [ fig : number_density_g=1 ] ) . for the fitting functions we set @xmath145 which are chosen to be consistent with initial conditions for numerical results.,title="fig : " ]
in this paper , we have described the effect of interactions in theories where the particle production is caused by a time - varying background field .
the necessary formalism has been introduced and analytic and numerical calculations have been performed in a simple but illustrative model . as a general conclusion , we have shown that the amount of produced particles is related not only to the usual bogoliubov coefficients @xmath25 but also to the magnitude and nature of the interaction terms . in order to evaluate interaction effects and to compare their impact with a contribution coming from @xmath25
, we have considered a specific ( supersymmetric ) model which contains both massless fields @xmath146 ( @xmath63 ) and massive fields @xmath77 whose mass is varying in time due to a background field @xmath6 ( hence these fields have @xmath64 ) . as the result ,
we have found not only explicit results for massive particles but also quite interesting effects for massless particles , which are induced by interactions of massless fields with massive ones and , directly and indirectly , with the background field .
as illustrated by the fitting functions , the massless distribution is suppressed by a factor about @xmath142 compared with the usual bosonic / fermionic distributions .
since these results are derived by the perturbative method , they are not necessarily valid in the truly strong coupling region .
however , we can safely conclude that it is possible to create a sizeable amount of particles with the help of interaction effects if the coupling is reasonably strong .
+ * acknowledgements * + this work has been supported by the polish ncn grant dec-2012/04/a / st2/00099 .
in the fermionic deliberations we use the eigenvector of the helicity operator @xmath147 ( @xmath148 ) satisfying the relation @xmath149 particular representation of this eigenvector is given by @xmath150 which implies the following relations : @xmath151 in this paper we choose orthogonality conditions of the form @xmath152 where @xmath153 is a phase factor defined as @xmath154 e. brezin and c. itzykson , phys .
d 2 , 1191 ( 1970 ) l. parker , phys .
183 , 1057 ( 1969 ) l. h. ford , phys .
d 35 , 2955 ( 1987 ) l. kofman , a. d. linde and a. a. starobinsky , phys .
73 , 3195 ( 1994 ) [ hep - th/9405187 ] l. kofman , a. d. linde and a. a. starobinsky , phys .
d 56 , 3258 ( 1997 ) [ hep - ph/9704452 ] l. kofman , a. d. linde , x. liu , a. maloney , l. mcallister and e. silverstein , jhep 0405 , 030 ( 2004 ) [ hep - th/0403001 ] s. enomoto , s. iida , n. maekawa and t. matsuda , jhep * 1401 * ( 2014 ) 141 [ arxiv:1310.4751 [ hep - ph ] ] .
c. n. yang and d. feldman , phys . rev .
79 , 972 ( 1950 ) e. t. akhmedov , n. astrakhantsev and f. k. popov , jhep * 1409 * , 071 ( 2014 ) [ arxiv:1405.5285 [ hep - th ] ] .
e. t. akhmedov and f. k. popov , arxiv:1412.1554 [ hep - th ] . | we have investigated effects of interaction terms on non - perturbative particle production .
it is well known that time - varying masses induce abundant particle production . in this paper
we have shown that it is possible to induce particle production even if the mass term of a particle species is not varying in time .
such particles are produced through the interactions with other fields , whose mass terms are varying due to a time - dependent background .
the necessary formalism has been introduced and analytic and numerical calculations have been performed in a simple but illustrative model .
the rather general result is that the amount of produced particles without time - dependent masses can be comparable with the particle density produced directly by the varying background if the strength of interaction terms is reasonably large .
* influence of interactions on particle production induced by time - varying mass terms * seishi enomoto , olga fuksiska and zygmunt lalak |
over the past 10 years , physicists have made a significant effort to understand unusual critical phenomena in complex network systems @xcite .
most of the real - world complex networks have been revealed as scale - free ( sf ) in the degree distribution , @xmath8 , where degree @xmath9 is the number of edges connected to a given vertex and @xmath1 is the degree exponent @xcite .
the critical phenomena of spin systems on such sf networks have been expected differ from those in euclidean space . according to various studies @xcite , the critical behavior of the ferromagnetic ising model on sf networks has a strong dependence on the degree distribution .
a spin - glass model has also demonstrated anomalous critical phenomena on sf networks .
the previous analytical study of the ising spin - glass model on sf networks @xcite showed that the phase diagrams consisting of the paramagnetic ( pm ) , ferromagnetic ( fm ) , and spin - glass ( sg ) phases were strongly dependent on the given degree distribution , and the transition temperature @xmath6 ( @xmath10 ) between the pm - sg ( pm - fm ) phases became infinite for @xmath7 .
then , provided an sg model based on the non - ising type is now considered in sf networks , how will its critical phenomena be changed ?
until now , little research has been conducted to answer this interesting question .
the main purpose of this paper is to investigate critical phenomena in sf networks by introducing a new non - ising - type sg model . for this purpose , we consider the ghatak - sherrington ( gs ) model @xcite , a spin-1 sg model with a crystal field .
recently , the gs model has become well known as a prototypical model for inverse transition @xcite .
inverse transition ( melting or freezing ) occurs when an ordered phase is more entropic than a disordered one . as a result
, the ordered phase appears at a higher temperature than the disordered one .
such inverse transitions have been experimentally observed in various physical systems , such as polymers @xcite , high-@xmath10 superconductors @xcite , magnetic thin films @xcite , and organic monolayers @xcite . the gs model in euclidean space shows
inverse freezing , wherein the sg phase becomes the one with higher entropy . nowadays
, the gs and analogous models with a crystal field have drawn attention by the observation of a bose glass " of field - induced magnetic quasiparticles in a doped quantum magnet @xcite .
in addition , erichsen _ et al . _ obtained the analytical solutions of the gs model with finite connectivity and found inverse transitions @xcite by extending the techniques of the previous ising spin - glass model with finite connectivity @xcite .
meanwhile , a quantum version of the gs model with a transverse tunneling field exhibited a peculiar phenomenon of the splitting within one sg phase , depending on the values of crystal and transverse fields @xcite . in this work
, we will study the gs - based sg model on sf networks in order to obtain its analytical solutions , i.e. , free energy and order parameters , and various phase diagrams . from these solutions
, we can expect two simultaneous unique features when an alternative sg model , based on the gs model , is studied in sf networks .
one such feature is that the alternative model may contain an inverse transition similar to that of the gs model in euclidean space .
the other is that phase diagrams and transition temperatures obtained from the alternative model may exhibit a new dependence of the degree distribution of the considered sf network .
however , the divergence of transition temperatures , shown in the ising spin - glass model on sf networks with @xmath7 , may disappear in the alternative model under the same degree distributions .
the following analysis on the alternative model will give clear answers about such expectations .
now , we will investigate the critical phenomena by considering the gs model on sf networks .
the sf networks , on which the gs model is placed , are constructed by the static model .
based on the replica - symmetric solution , we thereby obtain the analytical solutions , i.e. , free energy and order parameters . in order to construct sf networks , on which spin system is placed ,
we follow the process of the static model , where the name static " originates from the fact that the number of vertices is fixed from the beginning @xcite .
this model has the advantage for analytical calculations of its theoretical quantities , such as free energy and order parameters @xcite .
the general random graph under the static model is constructed as follows @xcite : the number of vertices @xmath11 is fixed at the beginning .
each vertex @xmath12 @xmath13 is assigned a weight @xmath14 .
a pair of vertices @xmath15 is chosen with the probabilities @xmath14 and @xmath16 , respectively , and they are connected with an edge , unless the pair is already connected .
this process is repeated @xmath17 times . in such random networks , the probability that a given pair of vertices @xmath15 ( @xmath18 ) is not connected by an edge , as denoted by @xmath19 , is given by @xmath20 , while the connection probability @xmath21 . for the erds - rnyi ( er ) graph @xcite , the weight is given as @xmath22 , independent of the index @xmath12 . since @xmath23 ,
the fraction of bonds present becomes @xmath24 and the average number of the connected edges is @xmath17 .
so @xmath2 becomes the mean degree in the er graph .
for the static sf network , the weights are given by @xmath25 where @xmath26 is a control parameter in the range @xmath27 , and @xmath28 . then the resulting network is an sf network with a power - law degree distribution , @xmath8 , with @xmath29 @xcite .
note that @xmath30 for finite @xmath2 , except @xmath31 for @xmath7 and @xmath32 . therefore , the mean degree of a vertex @xmath12 is @xmath33 and the mean degree of the network is @xmath2 @xcite . when @xmath2 approaches @xmath11 , this network becomes a fully - connected one or a regular lattice with infinite - range interaction .
therefore , we can say that the mean degree @xmath2 in this model plays a role similar to the physical dimension for spin systems on lattices .
now we consider an alternative sg model on static sf networks .
the sg transitions in euclidean space have already been studied by means of various theoretical methods @xcite .
most of such studies have concentrated on regular lattices or the infinite - range interaction model on fully connected graphs . to study the sg transitions on the static sf networks , we follow the previous approach of the dilute ising sg model with infinite - range interactions , i.e. , the ising sg model on the er graph , first performed by viana and bray @xcite , and applied successfully to the ising sg model on the static sf networks @xcite .
the hamiltonian of the gs model on a graph @xmath34 constructed by the static model is given as @xmath35 where @xmath36 is nonzero only when the vertices @xmath12 and @xmath37 are connected in @xmath34 .
for the static model , the probability of @xmath34 in the quenched random network ensemble is defined as @xmath38 with @xmath39 , @xmath14 being given in eq.(1 ) .
then , the ensemble average for a given physical quantity @xmath40 is taken as @xmath41 where @xmath42 denotes the average over different graph configurations . in the sg problem ,
the coupling strengths @xmath43 are also quenched random variables .
we assume that each @xmath36 is given as @xmath44 or @xmath45 with probabilities @xmath3 and @xmath46 , respectively , so the coupling strength distribution is given as @xmath47.\ ] ] the case of @xmath48 ( @xmath49 ) corresponds to the fully frustrated ( purely ferromagnetic ) case , and we consider @xmath3 in the range of @xmath50 throughout this work .
then the ensemble average for a given physical quantity , @xmath40 , is taken as @xmath51 where @xmath52 is an average over the quenched disorder of @xmath36 .
thus , the free energy @xmath53 is evaluated as @xmath54 , where @xmath55 is the partition function for a given distribution of @xmath43 on a particular graph @xmath34 . here
, the replica method is used to evaluate the free energy , i.e. , @xmath56/n$ ] . to proceed ,
we evaluate the @xmath57-th power of the partition function @xmath58 , @xmath59 , \label{z_n}\end{aligned}\ ] ] are the order parameters of the spin glass system , called the magnetization , the spin self - interaction , the spin - glass order parameter , and so on .
the average is evaluated through @xmath78 .
now we consider the replica - symmetric ( rs ) case , in which spins with different replica indices are indistinguishable , for simplicity .
furthermore , since order parameters such as @xmath79 are too complex to be obtained , for the present , we follow an approach that is similar in spirit to the sherrington - kirkpatrick model @xcite , in which terms that are higher in order than @xmath73 in eqs.(9 ) and ( 10 ) are neglected .
we thus determine the phase boundaries of the pm , fm , and sg phases through the rs solution with three order parameters : the rs magnetization , the rs spin self - interaction , and the rs sg order parameter are denoted as @xmath80 , @xmath81 , and @xmath82 , respectively , and the free energy in eq.(9 ) is truncated at the order of @xmath83 .
then we obtain the rs intensive free energy as follows : @xmath84.\ ] ] where @xmath85 , @xmath86 and @xmath87 .
we can determine @xmath88 , @xmath89 , and @xmath90 by imposing the condition that @xmath91 resumes the stable extrema when they are the replica - symmetric solutions .
from this extremal condition , we can obtain the self - consistent equations of @xmath88 , @xmath89 , and @xmath90 ( i.e. , @xmath92 ) as follows : @xmath93\\ r & = & \int \mathcal{d}z~ \sum_{i } p_{i}\bigg[\frac{2e^{\gamma_{i } } \cosh \eta_{i}(z)}{1 + 2e^{\gamma_{i } } \cosh \eta_{i}(z)}\bigg]\\ q & = & \int \mathcal{d}z~ \sum_{i } p_{i}\bigg[\frac{2e^{\gamma_{i } } \sinh \eta_{i}(z)}{1 + 2e^{\gamma_{i } } \cosh \eta_{i}(z ) } \bigg]^{2}\end{aligned}\ ] ] now , we consider three phases : pm ( @xmath94 ) , sg ( @xmath95 , @xmath96 ) , and fm ( @xmath97 , @xmath96 ) . under certain conditions
, there exists a multicritical point at which the pm - sg - fm phases merge .
in addition to the pm , sg , and fm phases , the mixed ( m ) phase is sometimes present in the present model .
the m phase is defined as the re - entrant sg phase with nonzero ferromagnetic order ( @xmath97 , @xmath96 ) , located below the fm phase @xcite .
the sg - m phase boundary is determined as the vertical straight line ( toulouse line ) from the multicritical point to @xmath98 @xcite .
the phase boundary between the replica symmetric phase ( pm , fm ) and the replica - symmetry - broken phase ( sg , m ) is determined by using the almeida - thouless ( at ) line @xcite , @xmath99 \equiv ( k \mathbf{t}_{2})^{-1 } - \int \mathcal{d}z \sum_{i } n p_{i}^2~ \bigg[\frac{4e^{2\gamma_{i } } ( 2e^{\gamma_{i } } + \cosh \eta_{i}(z))^{2 } } { ( 1 + 2e^{\gamma_{i } } \cosh \eta_{i}(z))^{4 } } \bigg ] = 0.\end{aligned}\ ] ] both the pm - sg phase boundary and the fm - m phase boundary are determined by the at line .
note that the rs solutions [ eqs .
( 14)-(16 ) ] are unstable below the at line , i.e. , inside the sg and m phases
. we can thus complete the phase diagrams of the present model from these equations .
we consider four phases : pm ( @xmath94 , @xmath100>0 $ ] ) , sg ( @xmath95 , @xmath96 , @xmath100<0 $ ] ) , fm ( @xmath97 , @xmath96 , @xmath100>0 $ ] ) , and m ( @xmath97 , @xmath96 , @xmath100<0 $ ] ) .
note that the er - type model can be obtained when we select the weight @xmath22 .
then our results ( @xmath91 , @xmath88 , @xmath89 , @xmath90 , and @xmath100 $ ] ) become simpler and independent of index @xmath12 . then the results appear mathematically identical to those of the gs model , except for the following main difference : instead of @xmath101 of the gs model , @xmath102 is used in the er - type model . the difference may be overcome when @xmath103 , i.e. , @xmath104 and @xmath105 .
therefore , in @xmath106 and @xmath107 , the main region of the inverse transitions , the gs model and the present model have very distinctive characteristics , in spite of the mathematical similarity of the results .
moreover , when @xmath14 depends on index @xmath12 in the sf model , the distinctiveness becomes larger .
the distribution of @xmath36 also differs between the two models : while the present model is based on the @xmath108 model , the gs model has a gaussian distribution of @xmath36 .
therefore , we may not obtain any meaningful information from the simple and direct comparison between the results of these two models .
the present model demonstrates a huge difference from the previous ising sg model @xcite , in that our model has no divergence of @xmath6 in the region @xmath7 : for the previous ising sg model @xcite , @xmath88 and @xmath90 approach zero near the pm - sg phase boundary , so the key term comprising free energy , @xmath109 $ ] in eq.(20 ) of ref .
@xcite , can be expanded by a series expansion of @xmath110 and @xmath111 , i.e. , the even multiples of small @xmath112 . according to ref .
@xcite , the term @xmath113 of the model was introduced by the series expansion , and the divergence of @xmath6 in the region @xmath7 occurred with the introduction of @xmath113 .
however , the present model does not allow for such a series expansion : @xmath114 in eq.(13 ) is always larger than 1 , although @xmath88 and @xmath90 approach zero near the pm - sg phase boundary . here
, the 1 " of @xmath114 results from @xmath5 , which is the differentiating spin value of the present model from the previous one .
thus , the present model has no room for the introduction of @xmath113 .
therefore , this model has no reason for any divergence of @xmath6 in the region @xmath7 .
@xmath6 in @xmath115 thus has a finite value , which is the main difference from the result of the previous model . in the previous subsection
, we neglected such terms that are higher in order than @xmath73 in eqs.(9 ) and ( 10 ) .
now we consider the perturbative approach @xcite , by which we expand the term of @xmath116 in eq.(9 ) up to fourth order . through this approach ,
we obtain the order parameters @xmath71 , @xmath117 , @xmath73 , @xmath74 , @xmath118 , @xmath76 and so on . for simplicity
, we use the notations defined by @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath123 , @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , and @xmath129 .
let @xmath130 represent a subset of the replica indices @xmath131 .
then it is convenient to denote the set @xmath132 as @xmath133 .
we also write @xmath134 . with these notations , @xmath135 where the sum is over all subsets of @xmath136 except the null set , and @xmath137 where @xmath138 .
this perturbative approach is to expand @xmath139 and keep only the terms up to fourth order .
using the properties that @xmath140 , @xmath141 for @xmath142 and so on , we rewrite the intensive free energy up to fourth order terms as @xmath143.\end{aligned}\ ] ] in this case , the 1 " of the @xmath144 $ ] term in eq.(19 ) results from @xmath145 . this simple effect by @xmath145 makes a huge difference with the previous ising sg model @xcite , as already pointed out in the previous subsection : provided the 1 " is absent , eq.(19 ) becomes similar to eq.(38 ) of ref .
@xcite , so order parameters @xmath146 can show singularities by divergences of @xmath6 ( or @xmath10 ) depending on the @xmath1 value .
however , the existence of the 1 " blocks any possibility of their singularity depending on @xmath1 value .
furthermore , terms having higher order @xmath147 , such as @xmath148 , @xmath149 and @xmath150 terms , in eq.(19 ) are smaller than the rest ones , so the contributions by higher order terms such as @xmath151 or @xmath152 may be negligible compared with those by @xmath153 and @xmath154 near the phase transition points in the thermodynamic limit .
thus , it becomes sufficiently meaningful to consider only the three rs order parameters @xmath88 , @xmath89 , and @xmath90 given in eqs .
( 14)-(16 ) .
from now onward , we consider the physical properties of the rs solutions and phase diagrams in detail .
for @xmath155 ( a ) , @xmath156 ( b ) , and @xmath157 ( c ) . here , @xmath158 , @xmath159 , and @xmath160.,title="fig:",scaledwidth=43.0% ] for @xmath155 ( a ) , @xmath156 ( b ) , and @xmath157 ( c ) . here , @xmath158 , @xmath159 , and @xmath160.,title="fig:",scaledwidth=43.0% ] for @xmath155 ( a ) , @xmath156 ( b ) , and @xmath157 ( c )
here , @xmath158 , @xmath159 , and @xmath160.,title="fig:",scaledwidth=43.0% ] figure 1 shows the dependence of @xmath90 on @xmath161 .
@xmath90 has a finite value of @xmath6 , even at @xmath162 .
as @xmath1 gradually increases , @xmath90 approaches an er case .
the @xmath163 case , as shown in fig .
1(a ) , shows no inverse freezing .
there is only the sg phase at any temperature in the range @xmath164 .
furthermore , @xmath6 approaches a finite value as @xmath115 .
it is well known that a sf network with a finite @xmath1 value over 2.0 has a few hubs holding together numerous vertices with small degrees , but the er network does not have such a hub .
therefore , as @xmath1 increases gradually from 2.0 , the number of hubs inside the sf network also decreases .
since the hubs have a large value of degrees , they tend to be frustrated by spins of numerous neighbor vertices .
since sg is defined as a complex system characterized by frustration , if the hubs under consideration have frustration , they play a crucial role in maintaining sg glass phase against any variation of physical parameters such as @xmath0 and @xmath161 .
as @xmath1 increases from 2.0 , @xmath6 thus decreases in proportion to the decrease of hubs . in fig .
1(b ) , an inverse transition occurs for all @xmath1 values when the @xmath161 value is nonzero .
the inverse transition has two critical temperatures , @xmath6 and @xmath165 , for the second - order p - sg and the first - order sg - p phase transitions .
the phase is varied in the order of pm @xmath166 sg @xmath167 pm , as the temperature is reduced .
this inverse transition thus corresponds to inverse freezing , according to the definition of schupper and shnerb @xcite .
inverse freezing depends on the value of @xmath1 : as @xmath1 increases from 2.0 , @xmath6 decreases but @xmath165 increases , because the range of the sg phase becomes narrower .
the er case with no hub has the narrowest region of the sg phase .
note that the phase transition at @xmath6 is of second - order , but the transition at @xmath165 is a first - order one . as shown in fig .
1(c ) , the larger value of @xmath161 reduces the difference between @xmath6 and @xmath165 . for @xmath168 and for the er case , no sg phase exists in the entire range of temperatures
the value of @xmath161 thus has a crucial effect on the inverse freezing of this model .
the effect of the @xmath161 field can be more clearly checked from phase diagrams shown in fig .
2 . phase diagrams for @xmath169 ( a ) and @xmath170 ( b ) . here
@xmath158 , @xmath159 , and @xmath160 . the solid - line ( dotted - line )
part of each phase boundary indicates the second - order ( first - order ) phase transition and each circle between the two kinds of lines denotes a tricritical point.,title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath169 ( a ) and @xmath170 ( b ) .
here @xmath158 , @xmath159 , and @xmath160 . the solid - line ( dotted - line )
part of each phase boundary indicates the second - order ( first - order ) phase transition and each circle between the two kinds of lines denotes a tricritical point.,title="fig:",scaledwidth=43.0% ] the graphs in fig .
2 show the @xmath171 phase diagrams obtained for specific @xmath1 values . for nonzero @xmath161 ,
inverse freezing occurs as the temperature is lowered .
even a very small value of @xmath161 can result in inverse freezing , by which the phase varies in the order of pm @xmath166 sg @xmath167 pm , as the temperature is reduced .
figure 2(a ) shows a tricritical point ( tcp ) , i.e. , the cross - point between first- and second - order phase boundaries , at @xmath172 . in the region of @xmath173 , therefore , only the pm phase exists . in fig .
2(b ) with @xmath170 , a larger value than that in fig .
2 ( a ) causes @xmath6 to decrease and @xmath165 to increase .
a tcp of this case is therefore located at the smaller value , @xmath174 , than that of fig .
the @xmath1 value thus plays an important role in the variation of @xmath6 and @xmath165 . such a role
is clearly shown in fig .
phase diagrams for @xmath175 ( a ) and @xmath176 ( b ) . here , @xmath158 , @xmath159 , and @xmath160 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath175 ( a ) and @xmath176 ( b ) . here , @xmath158 , @xmath159 , and @xmath160 .
, title="fig:",scaledwidth=43.0% ] the phase diagrams in fig .
3 show the dependence of @xmath6 and @xmath165 on @xmath1 . as shown in fig .
3(a ) , @xmath165 and @xmath6 under the nonzero @xmath161 value is also nonzero and finite , respectively , even at @xmath115 , which was already checked in fig .
1(b ) . figure 3(a ) clearly reveals that @xmath6 decreases and @xmath165 increases as @xmath1 increases from 2.0 .
the region of the sg phase thus becomes narrower as @xmath1 increases , but it does not disappear , even for a large value of @xmath1 . as shown in fig .
3(b ) , however , for sufficiently large @xmath161 , the @xmath177 phase diagram has a tcp at a specific @xmath1 value .
the case of @xmath176 shows the tcp at @xmath178 .
it is obvious that as the value of @xmath161 is increased , the tcp is located at a smaller value of @xmath1 .
phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath180 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath180 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath180 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath180 .
, title="fig:",scaledwidth=43.0% ] the graphs in fig . 4 show the @xmath181 phase diagrams obtained for @xmath169 . as mean degree @xmath2 increases ,
degrees of many vertices become larger . as a result
, the number of hubs also increases .
therefore , @xmath2 plays a fundamental role in the appearance of ordered phases like sg , fm , and m at nonzero @xmath0 . as shown in fig .
4(a ) , in the case of @xmath163 , the pm - sg transition occurs for @xmath2 values greater than a specific threshold ( @xmath182 ) , and there is no inverse transition . when @xmath161 becomes nonzero , as shown in fig .
4(b ) , a tcp is located at a certain threshold value ( @xmath183 for @xmath175 ) .
in addition , an inverse freezing , in the order of pm @xmath166 sg @xmath167 pm , occurs for @xmath184 as the temperature is lowered .
as @xmath2 gradually increases , @xmath6 also increases , however , @xmath165 decreases .
when only @xmath3 is increased under the same condition of fig .
4(a ) , as shown in fig .
4(c ) , fm and m phases as well as pm and sg phases occur , but no inverse transition exists . in fig .
4(c ) , the location of the multicritical point is ( 1.560 , 1.270 ) . when @xmath161 is increased under the condition of fig .
4(c ) , a very peculiar inverse transition occurs , as shown in fig .
4(d ) . in fig .
4(d ) , no sg phase exists , but there is a complex inverse transition , in the order of pm @xmath166 fm @xmath166 m @xmath167 pm , which occurs in the range of @xmath185 , as the temperature is reduced . furthermore , there are two tcps at @xmath186 and @xmath187 . it is uncertain whether this inverse transition can be called an inverse freezing or an inverse melting .
in addition , an inverse transition , in the order of pm @xmath166 fm @xmath167 pm , occurs for @xmath188 , as the temperature is reduced .
this inverse transition corresponds to an inverse melting , according to the definition of schupper and shnerb @xcite .
therefore , we checked that two alternative types of inverse transition , in addition to inverse freezing , can be discovered by controlling @xmath2 using proper values of @xmath3 and @xmath161 .
phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath168 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath168 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) . here , @xmath158 and @xmath168 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath159 and @xmath163 ( a ) , @xmath159 and @xmath175 ( b ) , @xmath179 and @xmath163 ( c ) , and @xmath179 and @xmath175 ( d ) .
here , @xmath158 and @xmath168 .
, title="fig:",scaledwidth=43.0% ] the graphs in fig . 5 show the @xmath181 phase diagrams obtained for @xmath170 , which are given for comparison with fig .
4 . figure 5(a ) shows a similar result to fig .
4(a ) , except for the threshold value ( @xmath189 ) . figure 5(b ) is also similar to fig .
4(b ) , but the threshold value , at which a tcp is located , is increased ( @xmath190 ) . while fig .
5(c ) is also similar to fig .
4(c ) , phase boundaries are shifted to a larger value of @xmath2 .
the location of the multicritical point is also shifted to ( 3.020 , 1.070 ) . in fig .
5(d ) , the m phase disappears for @xmath170 and an inverse melting , in the order of pm @xmath166 fm @xmath167 pm , occurs for @xmath191 , as the temperature is reduced . phase diagrams for @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath176 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath176 ( d ) . here , @xmath158 and @xmath160 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath176 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath176 ( d ) . here , @xmath158 and @xmath160 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath176 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath176 ( d ) . here , @xmath158 and @xmath160 .
, title="fig:",scaledwidth=43.0% ] phase diagrams for @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath176 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath176 ( d ) . here , @xmath158 and @xmath160 . ,
title="fig:",scaledwidth=43.0% ] figure 6 shows the @xmath192 phase diagrams obtained for several conditions . since @xmath3 has a value between fully frustrated case ( @xmath48 ) and purely ferromagnetic one ( @xmath49 ) , the increase of @xmath3 plays a role in replacing sg phase by fm or m phase . in fig .
6(a ) with @xmath180 and @xmath163 , the pm - sg transition occurs when @xmath3 is below 0.648 . in the region of @xmath193
, successive transitions occur in the order of pm @xmath166 fm @xmath166 m , as the temperature is lowered .
for @xmath194 , there is only the pm - fm transition . in fig .
6(a ) , the location of the multicritical point is ( 0.648 , 2.610 ) . when @xmath161 is increased to 1.5 , as shown in fig .
6(b ) , the inverse transitions occur throughout the range of @xmath3 .
when @xmath3 is smaller than 0.648 , inverse freezing occurs in the order of pm @xmath166 sg @xmath167 pm , as the temperature is reduced . as can be checked through the comparison between figs .
6(a ) and 6(b ) , the value @xmath195 , the location of the toulouse vertical line determining the sg - m transition does not depend on the @xmath161 value . instead
, the location of the multicritical point is shifted to ( 0.648 , 2.330 ) with @xmath176 . however , for @xmath196 , a complex inverse transition occurs in the order of pm @xmath166 fm @xmath166 m @xmath167 pm . moreover , there are two tcps at @xmath197 and @xmath198 . for @xmath199 ,
an inverse melting occurs in the order of pm @xmath166 fm @xmath167 pm , as the temperature is lowered .
therefore , for sufficiently large @xmath161 , we can find not only inverse freezing but also other inverse transitions by properly controlling the value of @xmath3 .
when only @xmath1 is increased under the same condition of fig .
6(a ) , fig .
6(c ) represents a similar result to fig .
6(a ) , but @xmath6 is lowered and the location of the toulouse line is shifted in the direction of large @xmath3 .
the location of the multicritical point is thus shifted to ( 0.707 , 1.550 ) .
when @xmath161 is increased under the condition of fig .
6(c ) , only pm and fm phases remain .
the tcp is located at @xmath179 , and for larger @xmath3 values , inverse melting occurs in the order of pm @xmath166 fm @xmath167 pm , as the temperature is reduced .
( @xmath165 ) vs. @xmath11 at @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath175 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath175 ( d ) . here , @xmath159 and @xmath160 .
in the figures , the blue diamonds and red rectangles indicate @xmath6 and @xmath165 , respectively.,title="fig:",scaledwidth=43.0% ] ( @xmath165 ) vs. @xmath11 at @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath175 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath175 ( d ) . here , @xmath159 and @xmath160 .
in the figures , the blue diamonds and red rectangles indicate @xmath6 and @xmath165 , respectively.,title="fig:",scaledwidth=43.0% ] ( @xmath165 ) vs. @xmath11 at @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath175 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath175 ( d ) . here , @xmath159 and @xmath160 .
in the figures , the blue diamonds and red rectangles indicate @xmath6 and @xmath165 , respectively.,title="fig:",scaledwidth=43.0% ] ( @xmath165 ) vs. @xmath11 at @xmath180 and @xmath163 ( a ) , @xmath180 and @xmath175 ( b ) , @xmath168 and @xmath163 ( c ) , and @xmath168 and @xmath175 ( d ) . here , @xmath159 and @xmath160 .
in the figures , the blue diamonds and red rectangles indicate @xmath6 and @xmath165 , respectively.,title="fig:",scaledwidth=43.0% ] all the figures we have checked were drawn for @xmath200 .
however , our previous theoretical results were obtained under the assumption of @xmath65 .
therefore , we should agree that the value of @xmath200 is small when compared with the assumption . then
, we must investigate whether such referred characteristics of each figure are still satisfied , even for larger @xmath11 values than 100 .
figure 7 shows the dependence of @xmath6 and @xmath165 on @xmath11 .
figure 7(a ) clearly shows that @xmath6 depends on @xmath11 for @xmath180 . according to fitting
, we check that @xmath201 . in fig.7(b ) with @xmath175 , inverse freezing by @xmath6 and @xmath165 exists , satisfying the fitting of @xmath202 and @xmath203 , respectively .
however , for @xmath168 , @xmath6 and @xmath165 show little dependence on @xmath11 , irrespective of @xmath161 values [ @xmath163 in fig .
7(c ) and @xmath175 in fig .
such little dependence seems to be natural , because @xmath88 , @xmath89 , and @xmath90 are independent of @xmath11 in the er case ( @xmath204 ) . therefore , fig .
7 represents all the results of previous figures that are quantitatively changed by the increase of the @xmath11 value , i.e. , @xmath6 increases but @xmath165 decreases by an increase of @xmath11 , and the changes become more sensitive as @xmath1 approaches 2.0 . since we can obtain the quantitative relation between @xmath6 ( or other transition temperatures ) and @xmath11 through the fitting method used in fig .
7 , we can draw the phase diagrams for all possible @xmath11 networks .
therefore , we can say that our previous results obtained from @xmath200 are still practically useful , even for the analysis of networks with very large @xmath11 .
we have studied the inverse transitions on sf networks through the static model . as already proven in ref .
@xcite , the static model enables one to study the sg problem by generalizing the dilute ising spin - glass model with infinite - range interactions to a model with inhomogeneous vertex weights .
we could also obtain the phase diagrams consisting of pm , fm , sg , and m phases , as functions of temperature @xmath0 , the degree exponent @xmath1 , the mean degree @xmath2 , and the fraction of the ferromagnetic interactions @xmath3 . however , the present model is based on the gs model , which considers the three states , including the @xmath5 state , contrary to the ising - based model , which considers only @xmath205 .
the @xmath5 state makes an essential feature when compared with the previous model @xcite : @xmath6 has a finite value even when @xmath7 .
this feature is opposite to the previous model , in which @xmath6 is infinite for @xmath7 .
furthermore , when crystal field @xmath161 has nonzero values , the present model shows three types of inverse transitions under specific conditions of parameters @xmath1 , @xmath3 , and @xmath2 .
the inverse freezing occurs in the order of pm @xmath166 sg @xmath167 pm , as the temperature is reduced .
the inverse melting occurs in the order of pm @xmath166 fm @xmath167 pm .
the third case can also occur in the order of pm @xmath166 fm @xmath166 m @xmath167 pm .
therefore , these two main results of the present model have special features as follows : until now , there have been few network models which do not contain the divergence of transitions for @xmath7 .
the present model may thus be considered as an original one to reveal critical behavior by including the third spin state ( @xmath5 ) .
in addition , the present model has a merit in that three types of inverse transitions can be simultaneously investigated by one type of model .
our investigations could be helpful in the understanding of various patterns in real systems with competing interactions , such as social networks .
usually , ising - like two - spin states may be considered to represent two different opinions in a society . in a similar way
, we can apply the present model with three - spin states to a real - world three - state systems .
one good example is the presidential elections carried out in many countries . when there are two candidates running for election , one segment of the people vote for a candidate in the ruling party , while others vote for a candidate in the opposition party , and the remaining portion abstains from voting in the election .
the existence of such abstentions is given by floating voters , who show frustration in deciding between the two candidates .
since sg is characterized by frustration , the sg phase can be a good analogy for a frustrated state in which the majority of people are floating voters , in spite of their firm will to vote .
similarly , the pm phase can express a state with nearly the same number of supporters for the two candidates , the fm phase is a state with a superiority of one party , and the m phase is a complex state , including both a little superiority of one party and floating voters .
the result of the election , i.e. , selection of the phase , can be determined by several external conditions , such as a degree of election fever ( @xmath0 ) , a degree of social mood toward indifference to the election ( @xmath161 ) , a relative approval rating by a public opinion poll ( @xmath3 ) , a degree exponent reflecting the network topology of the country holding the election ( @xmath1 ) , and a mean degree of people in the country ( @xmath2 ) .
when @xmath161 is zero , the sg phase by floating voters may be found even for the case of a very low election fever .
however , when @xmath161 is nonzero , the pm phase made by people who are indifferent to the election can be found for low election fever .
further research based on the data of real votes may prove the usefulness of the present model .
we believe that the methods we used here can be applied to other spin systems placed on sf networks .
moreover , we expect that the present model will help us extend our perspective of sg systems and of inverse transitions not only in theoretical models but also in real - world systems .
barabsi and r. albert , science * 286 * , 509 ( 1999 ) . r. albert and a. -l .
barabsi , rev .
phys . * 74 * , 47 ( 2002 ) .
s. n. dorogovtsev and j. f. f. mendes , adv .
phys . * 51 * , 1079 ( 2002 ) .
m. e. j. newman , siam review * 45 * , 167 ( 2003 ) .
a. aleksiejuk , j. a. holyst , and d. stauffer , physica a * 310 * , 260 ( 2002 )
. s. n. dorogovtsev , a. v. goltsev , and j. f. f. mendes , phys .
e * 66 * , 016104 ( 2002 ) .
m. leone , a. vzquez , a. vespignani , and r. zecchina , eur .
j. b * 28 * , 191 ( 2002 ) .
g. bianconi , phys . lett .
a * 303 * , 166 ( 2002 ) . c. p. herrero , phys .
e * 69 * , 067109 ( 2004 ) .
k. suchecki and j. a. holyst , phys .
e * 74 * , 011122 ( 2006 ) .
t. hasegawa and k. nemoto , phys .
e * 75 * , 026105 ( 2007 ) .
g. iannone and o. luongo , mod .
b * 25 * , 453 ( 2011 ) .
n. schupper and n. m. shnerb , phys .
lett . * 93 * , 037202 ( 2004 ) ; phys .
e * 72 * , 046107 ( 2005 ) .
a. crisanti and l. leuzzi , phys .
lett . * 95 * , 087201 ( 2005 ) .
l. leuzzi , philos . mag . * 87 * , 543 ( 2006 ) .
s. g. magalhaes , c. v. morais , and f. m. zimmer , phys .
b * 77 * , 134422 ( 2008 ) . c. v. morais , m. j. lazo , f. m. zimmer , and s. g. magalhaes , phys . rev .
e * 85 * , 031133 ( 2012 ) .
s. rastogi , g. w. h. hohne , and a. keller , macromolecules * 32 * , 8897 ( 1999 ) . c. chevillard and m. axelos , colloid polym .
sci . * 275 * , 537 ( 1997 ) ; j. desbrieres , m. a.v .
axelos , and m. rinaudo , polymer * 39 * , 6251 ( 1998 ) .
n. avraham , b. khaykovich , y. myasoedev , m. rapoport , h. shtrikman , d. e. feldman , t. tamegai , p. h. kes , m. li , m. konczykowski , k. van der beek , and e. zeldov , nature ( london ) * 411 * , 451 ( 2001 ) .
o. portmann , a. vaterlaus , and d. pescia , nature ( london ) * 422 * , 701 ( 2003 ) .
a. schll , l. kilian , y. zou , j. ziroff , s. hame , f. reinert , e. umbach , and r. h. fink , science * 329 * , 303 ( 2010 ) .
v. s. zapf , d. zocco , b. r. hansen , m. jaime , n. harrison , c. d. batista , m. kenzelmann , c. niedermayer , a. lacerda , and a. paduan - filho , phys .
* 96 * , 077204 ( 2006 ) .
r. yu , l. yin , n. s. sullivan , j. s. xia , c. huan , a. paduan - filho , n. f. oliveira jr , s. haas , a. steppke , c. f. miclea , f. weickert , r. movshovich , e. -d .
mun , b. l. scott , v. s. zapf , and t. roscilde , nature * 489 * , 379 ( 2012 ) .
r. erichsen jr . and
w. k. theumann , phys .
e * 83 * , 061126 ( 2011 ) .
r. erichsen jr . , w. k. theumann , and s. g. magalhaes , phys .
e * 87 * , 012139 ( 2013 ) . t. nikoletopoulos , a. c. c. coolen , i. prez castillo , n. s. skantzos , j. p. l. hatchett , and b. wemmenhove , j. phys .
a * 37 * , 6455 ( 2004 ) .
kim , phys .
e * 87 * , 042127 ( 2013 ) .
l. viana and a. j. bray , j. phys .
c * 18 * 3037 ( 1985 ) .
m. mzard and g. parisi , europhys . lett .
* 3 * , 1067 ( 1987 ) .
i. kanter and h. sompolinsky , phys .
* 58 * 164 ( 1987 ) .
p. mottishaw and c. de dominicis , j. phys .
a * 20 * l375 ( 1987 ) .
k.y . wong and d. sherrington , j. phys .
a * 21 * l459 ( 1988 ) .
r. monasson , j. phys .
a * 31 * 513 ( 1998 ) ; philos . mag .
b * 77 * 1515 ( 1998 ) .
k. h. fischer and j. a. hertz , _ spin glasses _
( cambridge university press , cambridge , 1991 ) .
j. a. mydosh , _ spin glasses : an experimental introduction _
( taylor & francis , london , 1993 ) . g. toulouse , j. phys .
( paris ) lett . *
41 * , l-447 ( 1980 ) .
j. r. l. de almeida and d. j. thouless , j. phys .
a * 11 * , 983(1978 ) . | in this paper , we will investigate critical phenomena by considering a model spin - glass on scale - free networks .
for this purpose , we consider the ghatak - sherrington ( gs ) model , a spin-1 spin - glass model with a crystal field , instead of the usual ising - type model .
scale - free networks on which the gs model is placed are constructed from the static model , in which the number of vertices is fixed from the beginning . on the basis of the replica - symmetric solution , we obtain the analytical solutions , i.e. , free energy and order parameters , and we derive the various phase diagrams consisting of the paramagnetic , ferromagnetic , and spin glass phases as functions of temperature @xmath0 , the degree exponent @xmath1 , the mean degree @xmath2 , and the fraction of the ferromagnetic interactions @xmath3 .
since the present model is based on the gs model , which considers the three states ( @xmath4 ) , the @xmath5 state plays a crucial role in the @xmath1-dependent critical behavior : glass transition temperature @xmath6 has a finite value , even when @xmath7 .
in addition , when the crystal field becomes nonzero , the present model clearly exhibits three types of inverse transitions , which occur when an ordered phase is more entropic than a disordered one . |
the relative entropy has been adopted as very important measure of statistical distinguishability . in the classical regime , the relative entropy of probability distribution @xmath2 to @xmath3
is defined by @xcite @xmath4 and also known as kullback - leibler divergence @xcite .
the usual convention is that @xmath5 , @xmath6 for @xmath7 .
the quantum relative entropy for ( normalized ) density operators @xmath8 and @xmath9 is defined as @xcite @xmath10 this expression is well - defined whenever the kernel of @xmath9 does not intersect with the support of @xmath8 , i.e. @xmath11 @xcite .
otherwise , the relative entropy is defined to be @xmath12 .
many fundamental results of quantum information theory are closely related to properties of the relative entropy @xcite .
there exist various generalizations of the standard entropic functionals .
the tsallis entropy has been found to be very useful in numerous problems of physics and other sciences @xcite . in the context of statistical physics ,
tsallis defined the @xmath0-entropy for @xmath13 by @xcite @xmath14 where the @xmath0-logarithm @xmath15 .
this functional has also been derived within an axiomatic approach @xcite .
entropic uncertainty principle has been expressed in terms of the tsallis entropies @xcite .
a generalization of the relative entropy for classical probability distributions is naturally introduced as @xcite @xmath16 note that the term ( [ dqdef ] ) can be recast as the sum @xmath17 , similar to the standard analytical form . in the quantum case ,
the sums in eq .
( [ dqdef ] ) are replaced by the traces of corresponding fractional powers of density operators . for @xmath18 ,
when no singularities occur , basic properties of the quantum relative @xmath0-entropy are examined in refs .
@xcite . up to a factor ,
the relative @xmath0-entropy is a particular case of petz s quasi - entropies @xcite . in the case @xmath1
, singularities may occur in the expression for the relative @xmath0-entropy . like the standard relative entropy
, the relative @xmath0-entropy is defined to be @xmath12 in the singular case .
relations between distinguishability measures are of interest . for certain applications , some of them may be more appropriate than others . for instance
, cloning processes are often studied with respect to fidelity - based measures @xcite , but the relative entropy has found use as well @xcite .
relations of such a kind are frequently expressed in the form of inequalities .
various upper bounds on the relative entropy ( [ relan ] ) were obtained in ref . @xcite . as given in terms of difference distances ,
these bounds characterize a continuity property in the sense of fannes @xcite .
recall that fannes inequality bounds from above a potential change of the von neumann entropy in terms of trace norm distance @xcite .
fannes inequality has been extended to the tsallis @xmath0-entropy @xcite and its partial sums @xcite . in the classical regime , continuity properties of wide classes of entropies and relative entropies
are considered in ref .
@xcite . in the present paper
, we are interested in upper continuity bounds on the relative @xmath0-entropy for @xmath1 .
in this section , the definition of the relative @xmath0-entropy for @xmath1 and related questions are discussed .
required mathematical tools are briefly outlined . by @xmath19
we denote the kernel of operator @xmath20 .
the support @xmath21 is the subspace orthogonal to @xmath19 .
[ maindf ] for @xmath1 and any pair of normalized density operators @xmath8 and @xmath9 , the quantum relative @xmath0-entropy is defined by @xmath22 let @xmath23 and @xmath24 denote the spectra of @xmath8 and @xmath9 , and let @xmath25 and @xmath26 denote the related orthonormal bases .
let @xmath27 and @xmath28 be functions of scalar variable @xmath29 .
recall that the formula @xmath30 is regarded as the definition of @xmath31 when @xmath32 or @xmath33 is unbounded .
the trace in ( [ reendef ] ) is written as @xmath34 which is generally unbounded for @xmath1 and singular @xmath9 .
if @xmath11 , i.e. @xmath35 , then we correctly define this trace as the trace taken just over @xmath36 , i.e. the sign @xmath37 is interpreted as @xmath38 .
effectively , the sum with respect to @xmath39 is also restricted to the nonzero @xmath39 s .
the quantum relative @xmath0-entropy enjoys some properties similarly to the standard relative entropy ( [ relan ] ) .
in particular , the measure ( [ reendef ] ) is positive and vanishes only for @xmath40 .
it is also pseudoadditive for all @xmath1 .
these points can be established by a relevant modification of the reasons given for @xmath18 in ref .
one of most important properties of the standard relative entropy ( [ relan ] ) is its monotonicity under the action of stochastic maps .
the relative @xmath0-entropy is jointly convex and monotone for @xmath41 .
these properties follow from the results of refs .
@xcite . more details and conditions for equality can be found in the recent review @xcite .
let @xmath42 be the space of linear operators on @xmath43-dimensional hilbert space @xmath44 .
by @xmath45 and @xmath46 we denote the sets of positive and strictly positive operators respectively . for any @xmath47
, we put @xmath48 .
the eigenvalues of @xmath49 counted with their multiplicities are the singular values @xmath50 of @xmath20 @xcite . for @xmath51 , the schatten @xmath52-norm of operator @xmath20
is given by @xcite @xmath53 this family includes the trace norm @xmath54 for @xmath55 and the spectral norm @xmath56 for @xmath57 .
these norms and relations between them have found use in various questions of quantum information @xcite .
for each @xmath58 $ ] and @xmath59 , there holds ( see , e.g. , section 2.4 in @xcite ) @xmath60 since the schatten @xmath52-norms are non - increasing in @xmath52 , these norms satisfy submultiplicativity @xmath61 . combining eq .
( [ schnp ] ) for @xmath55 with @xmath62 , we obtain @xmath63 we will extensively use the integral representations of matrix fractional power based on the formulas @xmath64 the second integral follows from the first one by substituting @xmath65 .
let us briefly mention lower continuity bounds .
lower bounds on the relative @xmath0-entropy for @xmath66 can be derived from the inequalities @xmath67 where @xmath68 . the relations ( [ p1qin ] )
are actually proved in ref .
( note that the definition of relative entropy in ref .
@xcite differs from ( [ relan ] ) in the sign , and the formula ( [ p1qin ] ) is obtained as the formula ( 9 ) of that paper with the reversed sign . )
the quantum relative entropy obeys ( see , e.g. , theorem 1.15 in ref .
@xcite ) @xmath69 this is a quantum analog of the pinsker inequality from classical information theory .
other lower bounds on the relative entropy ( [ relan ] ) are presented in ref .
@xcite . by eq .
( [ p1qin ] ) , these lower bounds are all valid for @xmath70 with @xmath66 . due to this fact and unboundedness of relative @xmath0-entropy for @xmath1 , we focus on nontrivial upper bounds .
the first upper bound extends the derivation given for the relative entropy ( [ relan ] ) in ref .
@xcite ( see example 6.2.31 ) . [ fup ] if both the density operators @xmath8 and @xmath9 are strictly positive then for @xmath66 , @xmath71 where @xmath72 , @xmath73
. * proof .
* using the second integral from eq .
( [ rint ] ) with @xmath74 and the linearity of the trace , one gives @xmath75 here we used @xmath76 and put @xmath77 .
due to eq .
( [ xzty ] ) , the relation ( [ fubin ] ) leads to @xmath78 indeed , the eigenvalues of @xmath79 are equal to @xmath80 , @xmath81 , and we find @xmath82 in terms of @xmath83 .
putting @xmath84 , we further obtain @xmath85 from eq .
( [ schnp ] ) , we get @xmath86 . by @xmath87 , we have the first inequality in eq .
( [ fub1 ] ) . the second inequality in eq .
( [ fub1 ] ) holds due to the fact that the operator @xmath88 is traceless and , hence , @xmath89 ( see lemma 4 in ref .
@xcite ) . in view of @xmath90 , the relation ( [ imfub ] ) gives eq .
( [ fub2 ] ) as well .
the above arguments go through for all @xmath91 .
since the term ( [ rqsq ] ) is a continuous function of @xmath0 , the claim remains valid for @xmath92 .
@xmath93 the bounds ( [ fub1 ] ) and ( [ fub2 ] ) are linear in the corresponding distance between @xmath8 and @xmath9 and characterize a continuity of the relative @xmath0-entropy in the fannes sense .
the former is more appropriate , when the maximal eigenvalue @xmath94 of @xmath8 is unknown and replaced with one .
the latter is stronger for sufficiently small values of @xmath94 .
they both show that the @xmath70 increases no faster than @xmath95 as @xmath96 goes to zero .
however , we would be interested in a bound which involves only the minimal eigenvalue of @xmath9 . this can be obtained by a @xmath0-parametric extension of the quadratic upper bound from ref .
@xcite ( see theorem 2 therein ) .
here one auxiliary statement is required .
[ frcon ] for any two @xmath97 and @xmath98 , there holds @xmath99 * proof . * the left - hand side of eq .
( [ aulem ] ) can be recast as @xmath100 , where the function @xmath101 is operator concave for all @xmath102 .
the concavity leads to @xmath103 where @xmath104 and @xmath105 . due to the integral ( [ rint ] ) ,
we rewrite ( [ corcon ] ) as @xmath106 in the limit @xmath107 , the right - hand side of eq .
( [ drivt ] ) tends to the right - hand side of eq .
( [ aulem ] ) , which is herewith the frchet differential of @xmath108 at @xmath109 in the direction @xmath110 .
@xmath93 [ secup ] for @xmath66 , if @xmath11 then @xmath111 where @xmath72 , @xmath112 , @xmath113 . *
* since @xmath114 gives @xmath115 for all @xmath116 , the formula ( [ aulem ] ) and the trace linearity lead to @xmath117 the right - hand side of eq . ( [ subin ] ) can be rewritten as the sum of two terms .
( [ xzty ] ) and the submultiplicativity of the spectral norm , the first term is estimated from above by @xmath118 here we have inserted @xmath119 , since the eigenvalues of @xmath120 are equal to @xmath121 , and the latter is an increasing function of @xmath122 .
dividing the right - hand side of eq .
( [ subiw1 ] ) by @xmath123 , we obtain the first term of the right - hand side of eq .
( [ secbn ] ) . using eq .
( [ xzty ] ) and the submultiplicativity again , the second term of the right - hand side of eq .
( [ subin ] ) is estimated by @xmath124 dividing this by @xmath123 , we obtain the second term of the right - hand side of eq .
( [ secbn ] ) .
@xmath93 the bound ( [ secbn ] ) is not purely quadratic , but its second term dominant for small @xmath125 is just quadratic in a distance between @xmath8 and @xmath9 . for traceless @xmath77
, we have @xmath89 ( see lemma 4 in ref .
@xcite ) , whence @xmath126 the upper continuity bounds ( [ secbn ] ) and ( [ secbn22 ] ) involve the minimal eigenvalue only of @xmath9 and suit for singular @xmath8 .
these points are advantages of these bounds . at the same time
, we would like to find an upper bound with the dependence @xmath127 in view of the term @xmath128 in the definition of @xmath70 .
indeed , for the standard relative entropy ( [ relan ] ) we have the bound which is logarithmic in the minimal eigenvalue of @xmath9 @xcite .
in this section , we present upper continuity bounds that cover all the values @xmath1 . moreover , these bounds has the dominant term with a dependence @xmath127 in the minimal eigenvalue of @xmath9 .
we first consider the case of integer powers .
[ sbmn ] for integer @xmath129 and @xmath130 , there holds @xmath131 where @xmath132 .
* we shall proceed by induction . for @xmath133 ,
the claim is obvious . for @xmath134 , we write @xmath135 by means of the triangle inequality , eq .
( [ schnp ] ) and @xmath136 , we then obtain @xmath137 assuming eq .
( [ lemsm ] ) , we derive @xmath138 too .
@xmath93 [ subbnm ]
let @xmath139 be a submultiplicative norm and @xmath130 . for integer @xmath129 ,
there holds @xmath140 where @xmath141 .
the claim can be proved similarly to the statement of lemma [ sbmn ] .
using lemma [ sbmn ] , we now obtain the upper bound of a kind @xmath127 for any @xmath1 , including non - integer @xmath0 . by @xmath142 and @xmath143
we respectively denote the floor and ceiling of real @xmath0 .
the main result of this section is stated as follows .
[ b0q1 ] for arbitrary @xmath1 , if @xmath11 then @xmath144 where @xmath145 , @xmath113 . *
proof . * for non - integer @xmath1
, we write @xmath146 , @xmath147 , and the identity @xmath148 by the linearity of the trace and @xmath123 , we then have @xmath149 due to eqs .
( [ xzty ] ) and ( [ lemsm ] ) , the first trace in the right - hand side of eq .
( [ rsns1 ] ) is bounded from above by the quantity @xmath150 the second trace is merely @xmath151 in view of @xmath152 and the positivity of the relative @xmath153-entropy .
combining this fact with eqs .
( [ rsns1 ] ) and ( [ rsns2 ] ) then gives the claim for non - integer @xmath1 .
for @xmath154 , the right - hand side of eq .
( [ rsns2 ] ) is a continuous function of the parameter @xmath0 .
the expression for @xmath70 is continuous for all @xmath13 .
so the bound ( [ dqm1 ] ) remains valid when @xmath0 tends to an integer from below .
@xmath93 with respect to small @xmath125 , the right - hand side of eq .
( [ dqm1 ] ) has a dependence @xmath127 .
so the upper continuity bound ( [ dqm1 ] ) is stronger than the bounds of theorems [ fup ] and [ secup ] .
the upper bound ( [ dqm1 ] ) also involves @xmath155 and the trace norm distance between density operators . by a structure , the right - hand side of eq .
( [ dqm1 ] ) is in good agreement with the expression for the relative @xmath0-entropy .
another advantage of the bound ( [ dqm1 ] ) is that it covers the range @xmath156 as well . at the same time , the factor @xmath157 is not continuous in @xmath0 .
for very interesting values , @xmath66 , the relation ( [ rsns2 ] ) leads to the inequality @xmath158 with @xmath159 instead of @xmath160 . in general
, the upper continuity bound ( [ dqm2 ] ) seems to be more appropriate among the ones presented above for @xmath66 .
for sufficiently small @xmath161 , this bound is clearly tighter than ( [ secbn ] ) and ( [ secbn22 ] ) . on the other hand , the upper bounds ( [ secbn ] ) and ( [ secbn22 ] )
may be more sharpening in the sense of closeness of the states @xmath8 and @xmath9 , when @xmath161 is not very small . due to @xmath162 ,
all the above bounds can be rewritten with one instead of the maximal eigenvalue .
finally , we note that our methods could be used for estimating the modulus of the second trace in the right - hand side of eq .
( [ rsns1 ] ) .
we present one continuity bound of such a kind , though it was not required for the proof of theorem [ b0q1 ] .
[ abst ] let @xmath116 , @xmath163 and @xmath164 .
for @xmath165 , there holds @xmath166 where @xmath167 , @xmath168 . * proof . *
using the first integral representation from eq .
( [ rint ] ) and the properties of the trace , we have @xmath169 combining eqs .
( [ bas ] ) and ( [ sab ] ) with the identity @xmath170 and putting @xmath104 , we obtain @xmath171 using the cyclic property of the trace and eq .
( [ xzty ] ) , one gets @xmath172 the eigenvalues of @xmath173 are equal to @xmath174 , the eigenvalues of @xmath175 are equal to @xmath176 , whence @xmath177 due to these relations and @xmath178 , the inequality ( [ abd2 ] ) leads to @xmath179 by the first integral of eq .
( [ rint ] ) , this is equivalent to eq .
( [ ab10 ] ) . @xmath93
similar to the standard relative entropy , the tsallis relative @xmath0-entropy is generally unbounded for @xmath1 . hence upper bounds on this functional are of interest . in this paper , we have obtained several upper bounds on the relative @xmath0-entropy for @xmath1 .
these bounds are expressed in terms of norm distances between its two arguments and the minimal eigenvalue of the second argument .
the presented inequalities characterize the property of continuity of the relative @xmath0-entropy for states which are close in the trace norm sense .
they also estimate from above the rate of divergence of the relative @xmath0-entropy when the minimal eigenvalue of its second argument goes to zero .
the considered upper bounds can be regarded as some @xmath0-parametric extensions of those bounds that have been obtained in the literature for the standard relative entropy . to derive the results , we have extensively used general properties of the trace and spectral norms as well as the known integral representations of matrix fractional powers . | generalized entropies and relative entropies are the subject of active research . similar to the standard relative entropy , the relative @xmath0-entropy is generally unbounded for @xmath1 .
upper bounds on the quantum relative @xmath0-entropy in terms of norm distances between its arguments are obtained in finite - dimensional context .
these bounds characterize a continuity property in the sense of fannes . |
the pearcey integral was first evaluated numerically by pearcey @xcite in his investigation of the electromagnetic field near a cusp .
the integral appears also in optics @xcite , in the asymptotics of special functions @xcite , in probability theory @xcite , as the generating function of heat ( and hence hermite ) polynomials of order @xmath6 for @xmath7 @xcite .
it also falls into the category of functions considered by plya @xcite , that is functions with countably many zeros . for the numerical evaluation of the zeros of the pearcey integral
see for instance @xcite , this will be important since this zeros will correspond to the initial value of each function @xmath8 .
the main motivation in finding the zeros of the pearcey function , which solves the heat equation @xmath9 is due to the fact that the main building block used to construct the density of the first time that a wiener process hits a boundary @xmath10 , is to find a function @xmath10 such that @xmath11 for example , suppose there is a financial contract which will be activated if ever the price of an asset @xmath12 ( modelled as brownian motion ) reaches a prescribed boundary @xmath10 .
for instance , in figure [ fig007 ] the blue line represents the evolution of the price of @xmath13 , for @xmath14 $ ] , in turn the red line represents a boundary which activates a contract if it is ever reached . in particular , the barrier option is a contract of this type . for a more detailed exposition
see for instance @xcite .
, in turn , the solid line represents a boundary which will activate a contract the first time it is reached . in this example the contract was activated approximately at time @xmath15 .
this is a random time , since it would have been impossible to foresee the outcome.[fig007 ] , width=340,height=264 ] next , we note that for some constant @xmath16 , the function @xmath17 with @xmath18 , solves the heat equation ( [ heat ] ) .
this is true since @xmath19 in ( [ linear1 ] ) is a linear combination of the fundamental solution of ( [ heat ] ) and its first derivative with respect to the space variable @xmath20 .
it is clear that the function ( [ linear1 ] ) equals zero at @xmath21 .
hence , for any @xmath22 , and setting @xmath23 in ( [ linear1 ] ) we obtain @xmath24 we note that the right - hand side of this identity is in fact the density of the first time that a brownian motion hits a linear boundary @xcite . in practice these results are used for instance in ( a ) the valuation problem of financial assets , in particular in the valuation of _ barrier options _ [ see bjrk ( 2009 ) ] , ( b ) in the quantificaction of counterparty risk [ see davis and pistorious ( 2010 ) ] , and in general in physical problems .
the main contributions of this work are , on the one hand , finding the zeros of the pearcey integral . on the other , advancing in the direction of developing a rather simple and straightforward methodology to find explicit solutions of the time - varying boundary problem for the heat equation . in this
regard we note that there exist techniques to study the latter aforementioned problem in terms of solutions to integral equations @xcite .
we recall that solutions in terms of integral equations , in general can only be evaluated numerically . in turn
, our approach leads to solutions in terms of odes .
the paper is organized as follows . in section [ secairy ]
we introduce the airy function of order 4 .
next in section [ secpearcey ] we define the pearcey integral and describe its connection with the airy function of order 4 . in section [ secrayleigh ]
we derive a rayleigh - type equation , whose solution kills the pearcey function .
the techniques described in section [ secrayleigh ] are illustrated with examples in section [ secexamples ] . in section [ secasymptotic ]
we derive a function which asymptotically solves the moving boundary problem for the pearcey integral .
this approximation can be helpful in the numerical solution of the rayleigh equation .
we conclude in section [ conclusions ] , with some final remarks .
with respect to the zeros of fourier integrals , plya proved @xcite that all the zeros of @xmath25 are real and infinitely many for @xmath26 . in turn
, the generalized airy function @xmath27 of order 4 can be expressed as a solution of the following ode @xmath28 one can prove , for instance applying the fourier transform to ( [ ai ] ) and solving the resulting equation , that @xmath27 is a particular case of ( [ poleq ] ) when @xmath29 , namely @xmath30 furthermore function @xmath27 is symmetric , with countably many zeros in the real line hence oscillatory and tends to zero as it increases to @xmath31 , see figure [ fig00 ] .
regarding the zeros of ( [ airy4 ] ) , there exist asymptotic estimates which are derived by means of the method of steepest descent @xcite . ,
defined in ( [ airy4 ] ) .
we observe the function is even and its first zero is at @xmath32.,width=340,height=264 ]
paris @xcite analyzed the asymptotic behavior of @xmath33 which by rotation of the path of integration @xmath34 and use of jordan s lemma ( see @xcite ) can be expressed as @xmath35 with @xmath36 and @xmath37 . in particular , the pearcey integral , which solves ( [ heat ] ) , is the case @xmath38 .
more explicitly , we have the following .
@xcitethe pearcey integral is defined as @xmath39 in this work we study instead the following fourier integral @xmath40 because the zeros of @xmath4 , for @xmath18 , are expressed in terms of a continuously differentiable function @xmath10 , as opposed to ( [ p2 ] )
. see @xcite .
we observe that the function @xmath4 in ( [ pearcey ] ) is the convolution between the kernel of standard brownian motion and the generalized airy function of order 4 , in equation ( [ ai ] ) .
throughout this work , the @xmath41-th partial differentiation with respect to the space variable @xmath20 of any given function @xmath42 is denoted as @xmath43 . in this section
we find the function @xmath10 for which the pearcey function is zero for every @xmath2 .
the idea is to exploit , on the one hand , the differential form of the airy function of order 4 , defined in ( [ ai ] ) , and on the other to use the fact that the pearcey function solves the heat equation ( [ heat ] ) .
the main result is the following .
[ th1 ] suppose that @xmath4 is as in ( [ pearcey ] ) , @xmath27 solves ( [ ai ] ) , @xmath44 is such that @xmath45 , and @xmath10 is a solution to the following rayleigh - type ode @xmath46 ^ 3-\frac{1}{2}tf'(t)-\frac{1}{4}f(t),\end{aligned}\ ] ] with @xmath47 and @xmath48 $ ] .
then , for every @xmath2 , we have @xmath49 given that @xmath27 is as in ( [ airy4 ] ) , its fourier transform equals @xmath50 furthermore , if we apply the fourier transform directly to the ode ( [ ai ] ) we have that @xmath51 since this expression is already in fourier domain , we convolve the previous expression with the heat kernel as follows @xmath52 this in turn , and by direct application of the integration by parts formula , yields @xmath53 as well as @xmath54 after differentiation with respect to @xmath20 .
next given that there exists an @xmath10 , see plya @xcite , such that the following holds for all @xmath55 @xmath56 we differentiate ( leibniz integral rule ) @xmath4 with @xmath57 , defined in ( [ pearcey ] ) , with respect to @xmath55 to obtain @xmath58 which is equivalent to @xmath59 and @xmath60 after differentiation with respect to @xmath55 twice .
we note that equations ( [ ai41 ] ) and ( [ ai42 ] ) obtained from the airy differential equation ( [ ai ] ) , as well as ( [ hit ] ) and ( [ hit2 ] ) obtained from the heat equation , involve derivatives of @xmath4 up to order 4 .
what remains is to obtain ( [ odesol ] ) from these expressions . to this end , from ( [ ai41 ] ) and ( [ ai42 ] ) we first have @xmath61 next from ( [ hit ] ) and ( [ hit2 ] ) it follows that @xmath62 these identities yield @xmath63 this completes the proof of theorem [ th1 ] .
to illustrate theorem [ th1 ] we next present examples and numerical experiments .
[ hx ] one can show that at @xmath64 the following two identities hold for @xmath10 : @xmath65 and @xmath66 .
hence , from theorem [ th1 ] , we may plot the solution of ( [ odesol ] ) , see figure [ figa ] , for @xmath67 $ ] .
the ode was solved numerically using mathematica . ....
airy4[x _ ] : = 1/(2*pi)*nintegrate[exp[i*x*y - y^4/4 ] , { y , -infinity , infinity } ] airy41[x _ ] : = nintegrate[i exp[i x y - y^4/4]*y , { y , -infinity , infinity}]/ ( 2 \[pi ] ) airy42[x _ ] : = nintegrate[-exp[i x y - y^4/4]*y^2 , { y , -infinity , infinity}]/ ( 2 \[pi ] ) x1 = findroot[re[airy4[x ] ] , { x , 2.44 } ] x1 = x /.
x1 fp = -n[re[airy42[n[x1]]]]/(2*re[airy41[n[x1 ] ] ] ) s = ndsolve[{g''[x ] = = -(g[x]/4 ) - 1/2 x g'[x ] + 2 ( g'[x])^3 , g[0 ] = = x1 , g'[0 ] = = fp } , g , { x , 0 , 4 } , accuracygoal - > 20 , precisiongoal - > 10 , workingprecision - > 33 ] plot[evaluate[g[x ] /.
s ] , { x , 0 , 4 } , plotrange - > all ] .... to test the accuracy of the solution in example [ hx ] we may use the following code in mathematica in the interval @xmath67 $ ] . .... test[x _ ] : = g[x ] /.
s f0[t _ , x _ ] : = 1/(2*pi)*nintegrate [ exp[i*x*y - y^2*t/2 - y^4/4 ] , { y , -infinity , infinity } ] table[re[f0[i/100 , test[i/100][[1 ] ] ] ] , { i , 0 , 400 , 1 } ] .... ) with @xmath65 , and @xmath66 .
the graph was plotted using mathematica.,width=340,height=264 ] next , we present some further examples of the methodology discussed in the previous section .
[ exai ] the airy function of order 3 solves the following ode @xmath68 using the same argument as in the proof of theorem [ th1 ] it follows that @xmath69 this last expression evaluated at @xmath10 yields @xmath70 finally , from the previous expression and ( [ hit ] ) it follows that @xmath71 thus , for some constant @xmath72 @xmath73 see ( * ? ? ? * ) . for applications of the airy function in the first hitting time problem of brownian motion up to a quadratic function see @xcite .
given the following ode @xmath74 and following the same line of reasoning as in the proof of theorem [ th1 ] we have @xmath75 from ( [ hit ] ) we obtain @xmath76 it follows that @xmath77 or @xmath78 for example if @xmath79 the moving boundary problem of the heat equation associated with @xmath27 is solved .
given the following bessel ode @xmath80 similar calculations as in the previous examples yield @xmath81 [ derivativeairy ] the derivative of the airy function @xmath82 solves @xmath83 which yields @xmath84 alternatively , from example [ exai ] we also have that @xmath85 using the same arguments as those described in section [ secrayleigh ] leads to @xmath86 this is the abel equation of the second kind and its solution can be expressed in terms of the airy function of order 3 @xmath87 , as follows : @xmath88 see figure [ figb ] for a numerical example with @xmath89 .
, defined in example [ derivativeairy ] , and such that @xmath89 was plotted with mathematica.,width=340,height=264 ]
in this section we carry out an analysis in order to find an asymptotic solution to the moving boundary problem associated with the pearcey integral .
this result is useful when finding the numerical solution of the rayleigh equation ( [ odesol ] ) .
the main result of this section is the following .
suppose that @xmath4 is as in ( [ pearcey ] ) , @xmath44 is any zero of the airy function of order 3 , and @xmath90 then @xmath91 as @xmath92 . for brevity
let us just consider the term within the brackets in ( [ pearcey ] ) , _ i.e. _ , @xmath93 introduce a variable @xmath94 , @xmath95 set @xmath96 and rearrange terms to obtain @xmath97 to get rid of the _ heat _ ( or quadratic ) term note that @xmath98 that is , @xmath99\pm\frac{i}{3^{1/2}}t^{1/2 } u^3-\frac{u^4}{4}\right\}e^{-\frac{\alpha^4}{4}-\frac{1}{2}\alpha^2t - i\alpha y}.\end{aligned}\ ] ] next , if we choose @xmath100 , as in ( [ alpha ] ) , @xmath101+\frac{i}{3^{1/2}}t^{1/2 } u^3-\frac{u^4}{4}\right\}e^{\frac{5}{36}t^2+\sqrt{\frac{t}{3}}y},\end{aligned}\ ] ] and thus , from ( [ pearcey ] ) , @xmath102+i\sqrt{3 t } \frac{u^3}{3}-\frac{u^4}{4}}du.\end{aligned}\ ] ] now , let @xmath103 , @xmath104 , and @xmath105 , which yields @xmath106}{(3t)^{1/6}}+i \frac{z^3}{3}-\frac{z^4}{4(3t)^{2/3}}}dz,\end{aligned}\ ] ] or equivalently @xmath107}{(3t)^{1/6}}+i \frac{z^3}{3}-\frac{z^4}{4(3t)^{2/3}}}dz.\end{aligned}\ ] ] letting @xmath108 @xmath109^{3/2}+\xi(3t)^{1/6}\right)=\frac{1}{2\pi}\int_{-\infty}^\infty e^{iz\xi+i \frac{z^3}{3}-\frac{z^4}{4(3t)^{2/3}}}dz.\end{aligned}\ ] ] hence for arbitrary @xmath44 @xmath110^{3/2}+\xi(3t)^{1/6}\right)\right\}=\hbox{ai}(\xi).\end{aligned}\ ] ] in particular if @xmath44 is a zero of the airy function then @xmath110^{3/2}+\xi(3t)^{1/6}\right)\right\}&=&\hbox{ai}(\xi)\\ & = & 0.\end{aligned}\ ] ] for instance , if @xmath111 we have that @xmath112 . see figure [ fig8 ] .
next , using the results of this section and section [ secrayleigh ] , we present an algorithm which can be used to solve ( [ odesol ] ) in theorem [ th1 ] . for some root @xmath44 of the airy function ai , let @xmath113 be defined as in ( [ aprx ] )
. given an arbitrary time @xmath114 we may find a solution to equation ( [ odesol ] ) in the interval @xmath115 $ ] as follows .
@xmath116 $ , for some $ \varepsilon>0$. } \end{array } \right.\end{aligned}\ ] ] suppose we choose @xmath117 and @xmath118 . then the procedure is the following .... f1[t _ , x _ ] : = 1/(2*pi)*nintegrate[(i*y ) * exp[i*x*y - y^2/2*t - y^4/4 ] , { y , -infinity , infinity } ] f2[t _ , x _ ] : = 1/(2*pi)*nintegrate[(i*y)^2 * exp[i*x*y - y^2/2*t - y^4/4 ] , { y , -infinity , infinity } ] asy[t _ ] : = 2*(t/3)^(3/2 ) - airyaizero[1]*(3*t)^(1/6 ) x10 = findroot[re[f0[10 , t ] ] , { t , n[asy[10 ] ] } ] x10 = t /.
x10 fp10 = -re[f2[10 , x10]]/(2*re[f1[10 , x10 ] ] ) s10 = ndsolve[{g''[x ] = = -1/4*g[x ] + 2*(g'[x])^3 - 1/2*x*g'[x ] , g[10 ] = = x10 , g'[10 ] = = fp10 } , g , { x , 8 , 12 } , accuracygoal - > 20 , precisiongoal - > 10 , workingprecision - > 30 ] .... , the blue is as in ( @xmath119 ) , with @xmath111 .
the graph was plotted with mathematica.,width=340,height=264 ]
due to the stochastic and periodic nature of several economic variables , as for instance mexico s general cpi or the fruit and vegetable annual inflation and assuming @xmath120 is a random walk , these processes can be modelled as @xmath121 where the @xmath122 and @xmath123 represent respectively the amplitude and phase at a time given frequency @xmath124 . in turn a continuous time approximation of ( [ sine ]
) can be expressed in terms of the solution of an sde of the form since @xmath125 where functions @xmath126 and @xmath127 are respectively : @xmath128 and @xmath120 is a wiener process .
a reasonable set of questions that could be asked could be for instance : @xmath129 \hbox{what is probability that the general cpi will remain}\\ \hbox{between 3 and 4 percent until the end of 2017 ? } \end{aligned}\ ] ] as it turns out , to answer the previous questions it is necessary to understand the moving boundary problem of heat equation addressed in this work .
more specific examples is still work in progress .
in this work we find the zeros of the pearcey function , in terms of the solution of a rayleigh - type equation .
this goal is achieved by exploiting , on the one hand , the differential equation of an airy function of order 4 and on the other by using the fact that the pearcey function is a solution of the heat equation . as a by - product
we develop a methodology , using straightforward techniques , to solve the moving boundary problem of the heat equation in the case in which the convolving function is a generalized airy function .
we expect that the techniques described within can be used in the construction of densities of the first hitting time problem of brownian motion .
the scope and applicability to the latter problem is still work in progress .
xx berry , m. v. and klein , s. ( 1996 ) colored diffraction catastrophes , _ proc .
usa _ * 93 * , pp .
bjrk ( 2009 ) . _ arbitrage theory in continuous time _ , oxford .
davis , m.h.a . and
pistorius , m.r .
quantification of counterparty risk via bessel bridges .
working paper .
de lillo , s. and fokas , a. s. ( 2007 ) the dirichlet - to - neumann map for the heat equation on a moving boundary , _ inverse problems _ , * 23 * , pp .
hernandez - del - valle , g. ( 2012 ) on hitting times , bessel bridges and schrdinger s equation , _ bernoulli _ , * 19 * ( 5a ) , pp . 15591575 .
kaminski , d. and paris , r. b. ( 1999 ) on the zeros of the pearcey integral , _ j. of comput . and appl .
_ , * 107 * , pp .
karatzas , i. and shreve , s. e. ( 1991 ) _ brownian motion and stochastic calculus _ , 2nd ed . _ graduate texts in mathematics _ * 113*. new york : springer .
martin - lf , a. ( 1998 ) the final size of a nearly critical epidemic , and the first passage time of a wiener process to a parabolic barrier , _ j. appl
_ , * 35 * , pp . 671682 .
paris , r. b. ( 1991 ) the asymptotic behaviour of pearcey s integral for complex variables , _ proc .
. london _ * a 432 * , pp .
paris , r. b. ( 1994 ) a generalization of pearcey s integral,_siam j. math .
_ , * 25 * , pp.630645 .
pearcey , t. ( 1946 ) the structure of an electromagnetic field in the neighbourhood of a cusp of a caustic , _ phil . mag .
_ , * 37 * , pp . 311317 .
plya , g. ( 1927 ) ber trigonometrische integrale mit nur reellen nullstellen , _ j. reine angew . math .
_ , * 158 * , pp .
rosenbloom , p. c. and widder , d. v. ( 1959 ) expansions in terms of heat polynomials and associated function , _ trans .
_ * 92 * , pp . 220266 .
senouf , d. ( 1996 ) asymptotic and numerical approximations of the zeros of fourier integrals , _ siam j. math .
_ , * 27 * ( 4 ) , pp .
tracy , c. a. and widom , h. ( 2006 ) the pearcey process , _ commun .
phys . _ * 263 * , pp .
valle , o. and soares , m. ( 2004 ) _ airy functions and applications to physics _ , imperial college press , london . | in this work we find a sequence of functions @xmath0 at which the integral @xmath1 is identically zero for all @xmath2 , that is @xmath3 the function @xmath4 , after proper change of variables and rotation of the path of integration , is known as the pearcey integral or pearcey function , indistinctly .
we also show that each @xmath0 is expressed in terms of a second order non - linear ode , which turns out to be of the rayleigh - type .
furthermore the initial conditions , which uniquely determine each @xmath0 , depend on the zeros of an airy function of order 4 defined as @xmath5 as a byproduct of these facts , we develop a methodology to find a class of functions which solve the moving boundary problem of the heat equation . to this end , we make use of generalized airy functions , which in some particular cases fall within the category of functions with infinitely many real zeros , studied by plya . |
the pricing of contingent claims contracts in financial economics is often based on very restrictive assumptions about the time evolution of the underlying instrument @xcite . in recent years
researchers have endeavored to remove some of these restrictions by proposing more realistic models which would incorporate features found in real markets @xcite .
there is however a trade - off between analytic tractability and adherence to stylized facts observed from empirical financial time series .
the most attractive features of the usual black - scholes type of models is the possibility of obtaining closed , exact formulas for the premium of derivative securities , and to build a risk free replicating strategy .
such qualities are amply used in financial institutions which require fast calculation and tools to hedge risky assets
. however , the geometric brownian motion assumption of the black - scholes models ignores several empirically observed features of the real markets such as , volatility clusters , fat tails , scaling , occurrence of crashes , etc .
it is as yet unknown which stochastic process is responsible for the motion of risky assets , but physicists have taken some important steps in the right direction@xcite . in this work
we implement a microscopic agents - based model of market dynamics which gives rise to a quite complex and rich behaviour@xcite and whose output are macroscopic quantities such as price returns . by varying its parameters
the model exhibits market crashes , gaussian statistics and short ranged correlation , fat tailed returns and long range correlation .
this model retains the nontrivial opinion formation structure of the grand canonical minority game@xcite because it incorporates two new features .
the first one is to allow two categories of agents , producers ( who use the market for exchanging goods ) , and speculators ( whose aim is to profit from price fluctuations ) .
the second feature is that speculators might choose not to trade , and in this sense the model is similar to the grand canonical ensemble of statistical physics since the number of active traders is not constant .
we perform an analysis of the time series generated by this model in order to classify its dynamical behaviour .
we first test the data for unit roots and remove a simple kind of nonstationarity by taking differences wherever necessary .
the bds statistic @xcite , originated in the chaos literature , uses the correlation integral @xcite as the basis of a test for the hypothesis that the data is independent and identically distributed ( _ i.i.d . _ ) .
we apply this statistic as a model specification test by applying it to the residue of an arima , autoregressive integrated moving average , process ( although this might not remove all kinds of linearity ) @xcite .
if the null hypothesis is rejected then this is indication that the data is nonlinear .
another test based on surrogate data @xcite is used to confirm the nonlinearity of the model . since the alternative hypothesis for the bds procedure is not specified other tests have to be applied in order to determine whether the nonlinearity comes from a stochastic or deterministic mechanism .
such distinction is subtle and here we approach this question by computing more parameters , the correlation dimension and two other procedures which do not require embeddings : recurrence plots @xcite , and the lempel - ziv complexity ( lzc ) @xcite . in complement to the above procedures we implement a bayesian approach called cluster weighted modelling ( cwm ) @xcite in order to find further indication of determinism .
the paper is organized as follows . the market model and its trajectories are analysed in section [ sec:2 ] .
the time series analysis , including all statistical tests is discussed in section [ sec:3 ] .
a succinct presentation of the cwm and its results are found in section [ sec:4 ] .
analysis of the results obtained and a classification of the model evolution are given in section [ sec:5 ] while in the last section we summarize the main results and comment on future work .
inspired by the _
el farol problem _ proposed by arthur @xcite , the so called minority game ( mg ) model introduced by challet and zhang @xcite represents a fascinating toy - model for financial market .
now it is becoming a paradigm for complex adaptive systems , in which individual members or traders repeatedly compete to be in the winning group .
the game consists of @xmath0 agents that participate in the market buying or selling some kind of asset , e.g. stocks . at
any given time agent @xmath1 can take two possible actions @xmath2 , meaning buy or sell . those players whose bets fall in the minority group are the winners , i.e. , the sellers win if there is an excess of buyers , and vice versa . to determine the minority group we just consider the sign of the global action @xmath3 , so that if @xmath4 is positive the minority is the group of sellers ; in this case the majority of players expect asset prices to go up .
in other words , this dynamics follows the law of demand and supply . in the end of each turn
, the output is a single label , @xmath5 or @xmath6 , denoting the winning group at each time step .
this output is made available to all traders , and it is the only information they can use to make decisions in subsequent turns .
indeed , they store the @xmath7 most recent output of winners set @xmath8 . in this way a limited memory of length @xmath7 is assigned to the traders corresponding to the most recent history bit - string that traders use to make decisions for the next step . in order to decide what action to take
, agents use strategies .
a strategy is an object that processes the outcomes of the winning sets in the last @xmath7 bets and from this information it determines whether a given agent should buy or sell for the next turn .
when a tie results from the strategies , buying or selling is decided by coin tossing .
the memory defines @xmath9 possible past histories so the strategy of one agent can be viewed as a d - dimensional vector , whose elements can be 1 or -1 .
the space @xmath10 of strategies is an hypercube of dimension @xmath11 , and the total number of strategies in this space is @xmath12 . at the beginning of the game
each agent draws randomly a number @xmath13 of strategies from the space @xmath10 and keeps them forever . after each turn
, the traders assigns one ( virtual ) point to each of the strategies which would have predicted the correct outcome . along the game
the traders always choose the strategy with the highest score .
the mg is a very simple model , capable of exhibiting complex behaviour , like phase transitions between an information - efficient phase and information - inefficient phase , by just varying a control parameter @xmath14 , e.g. , the ratio between information complexity and number of strategies present in the game . , @xmath15 , @xmath16 and @xmath17 and ( b ) sp500 data ] more realistic features are reached with the grand canonical minority game@xcite . in this model
we define the price process in terms of excess demand @xmath4 , @xmath18 and introduce two kinds of agents .
* the first kind is called producers , who go to the market only for the purpose of exchanging goods ; they only have one strategy and in this sense they behave in a deterministic way with respect to @xmath19 .
the number of producers is @xmath20 . *
the other kind of agents are the speculators , who go to the market to make profit from price fluctuations . since they are endowed with at least two strategies during the game they need to use the best strategy ; in this sense speculators are adaptative agents with bounded rationality .
the number of speculators is @xmath21 . in this version of the game ,
the number of speculators can change anytime , because the agent may decide not to trade in which case @xmath22 . since the strategy chosen by speculators is the one with highest score
, it is very important to update the scores of each strategy at each time step .
the updating of the scoring @xmath23 of each strategy @xmath24 belonging to agent @xmath1 is given by the following equations @xmath25 where @xmath26 is a threshold parameter .
a sample of the trajectory generated by this model is shown in fig .
[ fig : return](a ) .
an important piece of information to understand the generating mechanism of the system can be obtained from the periods where no coin is tossed .
the game is necessarily deterministic during this mode of evolution and we selected a rather long period of 1114 points to perform a time series analysis in this regime .
this work is based upon the grand canonical mg model , henceforth called mg model .
in this section we discuss the tools employed in time series analysis starting with the bds statistics .
we generate @xmath27 points for each of the benchmark data described in the appendix .
another benchmark is the set comprising @xmath28 closing prices for the index sp500 from january 01 , 1965 to january 01 , 1995 , shown in fig .
[ fig : return](b ) . all time series used here are unit root stationary with the exception of the financial index : this nonstationary is removed by taking first difference .
the bds test uses the correlation integral @xcite as the basis of a statistic to test whether a series of data is _
_ in the chaos literature the correlation integral is part of an efficient tool to compute the fractal dimension of objects called attractors ( for a formal definition of attractor see , e.g. , ref .
@xcite ) . given a sample of empirical data @xmath29 ,
the theory of state - space reconstruction @xcite requires that the @xmath30-histories of the data be constructed , where @xmath30 is called the embedding dimension . under certain conditions it is possible to reproduce in this space the dynamics of the system for a correct choice of @xmath30 .
the correlation integral is a function defined on the trajectories in this space and from it one can compute the correlation dimension . a simple test for determinism consists of increasing the embedding dimension and observing the occurrence of a corresponding increase in the correlation dimension .
some conditions have to be met in order to apply this method , mainly stationarity and sufficient number of data points @xcite .
our results show that the increase of the correlation dimension for the mg model with respect to the embedding dimension is practically identical to the stochastic benchmark series , including the index sp500 . in this work
we resort to better ways of analysing the occurrence of stochastic behaviour in complex time series evolution .
the bds statistic will be applied to the residue of arima processes in order to detect nonlinearity in the data . in this sense
the statistic is used as a specification test .
the asymptotic distribution of the statistic under the null of pure whiteness , is the standard normal distribution .
the alternative hypothesis is not specified @xcite .
the code implemented here is taken from ref .
@xcite . from table
[ table1 ] , the null hypothesis for the mg model and for the sp500 index are rejected more strongly than for the nonlinear stochastic model @xmath31 . in the surrogate data analysis
, a null hypothesis is tested under a measure @xmath32 , usually some nonlinear statistic .
surrogates are copies of the original time series preserving all its linear stochastic structure .
let @xmath33 and @xmath34 be the test measure for the original and the surrogate data sets , respectively .
the null hypothesis is rejected only if @xmath35 or @xmath36 for all surrogates .
some care should be taken to test the null hypothesis , otherwise false - positive rejection of null hypothesis could result .
a detailed analysis on this is found in @xcite . here
surrogate data sets @xcite are used in complement to the bds statistic to test for nonlinearity .
the null hypothesis is that the data is described by a stationary linear stochastic process with gaussian inputs .
the test statistic @xmath32 is a simple measure of predictability as in @xcite and the procedure is to generate @xmath37 samples of rescaled surrogates and to compare with the mg model .
if the test statistic falls outside the interval defined by the surrogates , then the null is rejected at the @xmath38% level .
we found that , for embedding dimensions from 2 to 5 , the mg is nonlinear at this level of significance , corroborating the previous result using the bds . in the remaining of this section
we discuss methods which do not require embeddings .
another measure of randomness that provides further insight into time series dynamics is the lempel - ziv complexity @xcite .
no embedding is necessary and the data is interpreted as a binary signal generated by some kind of source .
this idea is ever present in communication theory where one wishes to determine the minimum alphabet required to code a source whose signal is to be sent through a noisy channel . let us consider the length @xmath39 of the minimal program that reproduces a sequence with @xmath0 symbols .
the lempel - ziv algorithm is constructed by a special parsing which splits the sequence into words of the shortest length and that has not appeared previously .
for example , the sequence @xmath40 is parsed as @xmath41 .
one can show that @xmath42 where @xmath43 is the number of distinct words in a parsing and @xmath0 the size of the sequence . from this one
can see that @xmath39 contains a measure of randomness where a source that produces a greater number of new words is more random than a source producing a more repetitive pattern .
in analogy with dynamical evolution , those systems that are composed of well defined cycles are predictable while chaotic motion and stochastic processes are always producing new kinds of trajectories that never repeat themselves .
a comparison between chaos generated by differential equations and stochasticity can now be obtained . in the former case
the lempel - ziv complexity is well below 1 while in the latter it is close to this value .
more specifically , if we consider an oscillatory system such as the well known van der pol oscillator , then @xmath44 . for the lorenz attractor with 2000 points @xmath45 . in section [ sec:5 ] we comment on the discrepancy between this value and that in table [ table2 ] , and discuss the use of this complexity measure in dynamical systems generated by maps .
the complexity for the mg dynamics is found to be @xmath46 while the financial index has @xmath47 .
the implementation of recurrence plots @xcite used here is taken from @xcite since it provides a clear distinction of the systems we intend to classify .
the idea is to detect regions of close returns " in a data set .
the construction of the plots is simple : just compute the absolute values @xmath48 for all the points in the data base .
if the horizontal axis is designated by @xmath1 , corresponding to @xmath49 , and the vertical axis by @xmath50 , corresponding to @xmath51 , then plot a black dot at the site @xmath52 , whenever the absolute value difference is lower than @xmath53 ; otherwise plot it white .
actually the time series is normalized to @xmath54 and then @xmath53 is taken as @xmath55% of the average distance between successive points .
the black / white pattern can be used to detect determinism in the data .
there is a clear difference amongst plots generated by differential equations , maps , random data and stochastic processes as shown in fig .
[ fig : rp ] .
patterns of horizontal segments in the recurrence plot indicate the presence of unstable periodic orbits in maps or differential equations . in this sense
, these plots detect low dimensional chaos in relatively small and even noisy data sets .
an interesting probability density estimation approach to characterize and forecast time series developed by gershenfeld , schoner and metois @xcite is the so called cluster - weighted modelling .
this seems to be a powerful technique as it characterizes extremely well the time series of nonlinear , nonstationary , non - gaussian and discontinuous systems using probabilistic dependence of local models .
the cluster - weighted modelling technique estimates the functional dependence of time series in terms of delay coordinates .
the main task of this approach is to find the conditional forecast by estimating the joint probability density .
component is fitted by @xmath56 where @xmath57 , @xmath58 with delay @xmath59 and ( b ) minority game model , return is fitted in a similar fashion as in ( a ) ] let @xmath60 be the @xmath0 observations in which @xmath61 are known inputs and @xmath62 are the corresponding outputs . by knowing the joint probability density @xmath63 , we can derive the conditional forecast , the expectation value of @xmath64 given @xmath65 , @xmath56 .
we can also deduce other quantities such as the variance of the above estimation .
actually , the joint density @xmath63 is expanded terms of clusters which describe the local models .
each cluster contains three terms namely , the weight @xmath66 , the domain of influence in the input space @xmath67 , and finally the dependence in the output space @xmath68 .
thus the joint density can be written as @xcite , @xmath69 once the joint density is known the other quantities can be derived from @xmath63 . for example , the conditional forecast is given by , @xmath70 here @xmath71 describes the local relationship between @xmath65 and @xmath64 .
the parameters @xmath72 are found by maximizing the cluster - weighted log - likelihood .
the simplest approximation for the local model is with linear coefficients of the form , @xmath73 the method just described is capable of modelling a wide range of deterministic time series .
here we use cluster weighted modelling to distinguish between deterministic and stochastic time series . in deterministic systems
we observe that the variances of the different clusters converge to values lower than the variance of the original time series and one can verify that this property is robust under changes of the number of clusters . on the other hand stochastic systems do not have this property .
[ fig : cwm ] illustrates the comparison between the fitting of a deterministic system ( lorenz ) and the minority game data using cluster weighted modelling .
the main objective is to understand the minority game mode of evolution and other similar time series behaviour .
although this system is not generated by any kind of differentiable dynamics or even stochastic differential equations , we use in its analysis methods from dynamical systems , stochastic processes and complex systems theory .
tables [ table1 ] & [ table2 ] and fig .
[ fig : rp ] summarize the main results .
we will make frequent reference to the sp500 index since the minority game model is supposed to reproduce the dynamical evolution of financial markets and this index is used as a benchmark for comparison .
.bds statistic [ cols= " < , < , < , < " , ] the lempel - ziv complexity is an important parameter that can be used in the analysis of complex systems .
its advantage is that it does not require embeddings and can be easily employed in conjunction with other methods .
the results in table [ table1 ] supports the idea that there is a stochastic mechanism in operation driving the mg model .
there is a higher degree of indeterminacy in the sp500 and this is perhaps due to the fact that in this index there is a certain amount of measurement noise .
the surprisingly low complexity of the lorenz system is comparable to that of limit cycles , e.g. van der pol oscillator .
the explanation for this comes about when we compute its complexity for shorter time series .
for example at 2000 points the complexity is about @xmath74 times higher than the complexity for a 10000 length series as reported in table [ table2 ] .
this phenomenon does not occur for the discrete system like henon attractor .
due to the fact that the lorenz attractor contains a dense set of unstable periodic orbits , long time evolution affects the computation of the complexity and reveals some resemblance with periodic systems .
effects of this magnitude did not appear in the other time series analysed . in the deterministic intervals of the mg model
, the complexity has a value comparable to that of the chaotic henon map . the lempel - ziv complexity , as any other
test or statistic , should always be used in conjunction with other diagnostic tools . in all simulations
performed so far we have never found a stochastic process with complexity less than 0.8 .
however there are chaotic systems with complexity beyond this value , for example the family of maps @xmath75 , @xmath76 a prime number .
another result is that known chaotic systems described by differential equations do not have high complexity . to conclude that a system is stochastic we employ recurrence plots and the cluster weighted modelling approach .
the complexity is then used as a confirmation and , more importantly , to associate a level of stochasticity .
this is done in the same way as using lyapunov exponents to quantify chaos once determinism has been found .
the recurrence plot is a visual method which helps in the identification of similarities and differences amongst diverse modes of evolution .
several tests with differential equations and maps , represented here by the lorenz and henon systems , show that it is unlikely that low dimensional chaos can describe the kind of evolution found in the markets and in the mg model . in particular ,
recurrences are clearly identified in the lorenz system and nothing of the kind will ever appear in stochastic models . in this sense stochastic processes
are better suited to describe the financial index and the mg model .
however , in the deterministic regime one can identify close return patterns similar to those of low dimensional chaotic systems .
cluster weighted analysis reveals another aspect of the minority game behaviour .
its use in the modelling of general deterministic systems produces clusters whose variances are always smaller than the variance of the data . in contrast to this ,
when applied to a stochastic system the variances of the clusters are comparable to the variances of the data .
such distinction is preserved when we vary the number of clusters .
the issue of nonstationarity is a subtle one .
we limited ourselves in this study to unit root stationarity , but more sophisticated methods need to be used for the several brands of mg models and financial indices . using the complexity parameter we found that long time evolution reveals some intrinsic features of chaotic attractors described by differential equations . also , this parameter associated a higher complexity to the sp500 index as compared to the mg model and a possible explanation for this was given above .
the recurrence plots confirm that , in general , the mg model can not be described by low dimensional chaotic systems .
the nonlinear character of the model and the index are clearly indicated by the bds test and surrogate data analysis .
the mg model parameters employed in our simulations were chosen in the information efficient region . in the inefficient case
our simulations show that the system is a non random process with complexity close to zero .
the recurrence plot in this case provides a strong evidence of recurring orbits with several traces of horizontal segments .
thus , randomness is directly related to efficiency .
an important finding in our analysis of the mg game is that low dimensional chaotic regimes are possible during the evolution .
this occurs for intervals of about 1000 iterations , while shorter intervals of deterministic evolution are difficult to classify .
a complete study of the probabilistic structure of the deterministic phases , and their statistical significance , requires a more extensive investigation @xcite . for the full motion , comprised of transitions between deterministic modes induced by random perturbations ,
the time series analysis has indicated the operation of a nonlinear stochastic process which is similar , but not identical , to the sp500 index . in real markets
it is possible to find periods of deterministic behaviour in exchange rates for certain instruments and specific periods of time @xcite . in the sp500
such occurrence is unlikely but an extensive search in the data base for several lengths and starting points is feasible . even if deterministic modes can be found in historical stock prices , indices or rates , turning this into potential profit is a remote , but not an entirely discarded , possibility worthwhile pursuing .
another interest in this investigation is that it provides a test amongst several models . in this sense
one selects the models whose time series properties reproduce better real market evolution . in derivatives pricing a widely used test , and parameter estimation procedure ,
is to minimize the hedging error .
this can be implemented only if a model is available , e.g. black - scholes . in a non - gaussian context
there is no consensus on which model is appropriate .
an extension of this work , which will be pursued in a future work , is to use time series methods in the several brands of agents based models and to compare them with market data .
in addition to the stylized facts we should include nonlinearity , at least for the sp500 , and this is an additional reason to discard simple models based on the geometric brownian motion .
additional investigations in this study is to model the trajectories of the minority game model having in view the pricing of derivatives instruments , and to use a more extensive set of statistics to examine other financial series in order to confront the similarities and differences of this model with real market data .
the time series used as benchmarks were chosen to represent the kind of behaviour we intend to identify in the evolution of the minority game model . the lorenz system and the henon mapping are prototypes of deterministic behavior generated by differential equations and differentiable mappings .
arma models and the nlma are examples of linear and nonlinear stochastic processes . in the following
we describe briefly the models used in the present study .
nonlinear differential equations @xmath77 the parameter values are chosen as @xmath78 , @xmath79 and @xmath80 .
the data series is obtained by solving eqs .
( [ eq : lorenz ] ) numerically using fourth order runge - kutta method .
f. takens , detecting strange attractors in turbulence , in : d. rand , l .- s .
young ( eds . ) , dynamical systems and turbulence , vol .
898 of lecture notes in mathematics , springer - verlag , berlin , 1981 , pp . 366381 .
p. e. rapp , c. j. cellucci , t. a. a. watanabe , a. m. albano , t. i. schmah , surrogate data pathologies and the false - positive rejection of the null hypothesis , int .
. chaos 11 ( 2001 ) 983997 . | the minority game ( mg ) model introduced recently provides promising insights into the understanding of the evolution of prices , indices and rates in the financial markets . in this paper
we perform a time series analysis of the model employing tools from statistics , dynamical systems theory and stochastic processes . using benchmark systems and a financial index for comparison ,
several conclusions are obtained about the generating mechanism for this kind of evolution .
the motion is deterministic , driven by occasional random external perturbation .
when the interval between two successive perturbations is sufficiently large , one can find low dimensional chaos in this regime .
however , the full motion of the mg model is found to be similar to that of the first differences of the sp500 index : stochastic , nonlinear and ( unit root ) stationary . , , , and |
modern nucleosynthesis studies require large reaction networks , often including thousands of nuclei between the driplines and their respective reactions with light particles @xcite .
astrophysical reaction rates employed in reaction network calculations are determined either directly from cross sections or from the rate for the inverse reaction by applying detailed balance @xcite .
the cross sections are either known from experiment or predicted by theory . even when a reaction is experimentally accessible ,
often astrophysical rates can not be directly measured .
excited states are thermally populated in a stellar plasma whereas only reactions on the ground state of the target can be investigated in the laboratory . here
i want to address two points which are interesting in the study of astrophysically relevant reactions , both for experimentalists and for theorists .
the investigation of electromagnetic transitions at low energies is of high interest to the fields of both nuclear structure and nuclear reactions .
frequently , the question arises how astrophysical capture reactions are affected by assumed changes in the photon strength function ( or the nuclear level density ) .
photon transitions to a state at excitation energy @xmath2 considerably increase in strength with increasing energy @xmath3 , both for e1 and m1 . on the other hand , the nuclear level density @xmath4 is strongly increasing with increasing @xmath2 , i.e.decreasing @xmath5 . folding the two dependences gives rise to an energy window for the main contributions ( see fig . [
fig : window ] ) @xcite .
fractionation of the smooth behavior may occur at low excitation energy where only few , isolated excited states are available and the use of a level density function is not appropriate . however , far off stability there is no information on excited states and most predictions employ a level density just above the g.s .
sometimes this is , however , appropriate when excited states are known but transitions to them are suppressed by spin selection rules . then transitions in the region of higher level density will still contribute most .
it is instructive to derive the energy window of relevant transitions across the nuclear chart for neutron , proton , and @xmath6 capture at astrophysical projectile energies .
figs.[fig : neutron][fig : alpha ] show the location of the maximum contribution for 60 kev neutrons , 5 mev protons , and 10 mev @xmath6s on isotopic chains of different elements .
due to the higher projectile energies the maximum can lie above the particle separation energy for the charged particle reactions whereas it is always below or at the neutron separation energy for neutron captures .
interestingly , the maximal contribution mostly originates from @xmath0 transitions of @xmath7 mev , quite independent of target and reaction type .
exceptions are neutron capture reactions to final nuclei with low level density , either at magic numbers or close to the driplines . because the level density is low and the full energy window can not be covered by the low - energy neutrons in these cases , the maximal contribution is stemming from transitions close to the highest possible energy ( and allowed by spin selection ) .
an example for this is @xmath8sn where transitions with @xmath9 are contributing most . also neutron captures close to the neutron dripline exhibit similar behavior .
it has to be cautioned , however , that the statistical hauser - feshbach model of nuclear reactions is not applicable for nuclei with low level density @xcite and direct capture reactions with @xmath0 emission to the continuum dominate @xcite , for which the present considerations are not valid , anyway .
astrophysical reaction rates @xmath10 include transitions from thermally excited states @xmath11 of a nucleus whereas rates @xmath12 derived from laboratory cross sections only consider g.s
. transitions .
the importance of excited states is given by the stellar enhancement factor @xmath13 .
the thermal population follows a maxwell - boltzmann statistics @xmath14 .
additionally , each transition has a transition probability @xmath15 so that the total contribution of transitions from a given state is @xmath16 .
it is common to suggest to measure reactions of astrophysical importance in the direction of positive @xmath1 ( forward reaction ) because it is generally assumed that @xmath17 .
this is due to the fact that there are more transitions with higher relative energy @xmath18 ( and thus higher @xmath19 ) contributing in the reverse direction .
however , reactions having different coulomb barriers in the entrance and exit channels may show a converse behavior @xcite .
if the coulomb barrier is significantly higher in the reverse direction than in the forward direction , @xmath20 may become even smaller than @xmath21 because transitions with slightly lower relative energy are already suppressed ( even when possible within spin selection rules ) and only the g.s .
( or lowest few excited states ) contributes .
when calculating @xmath22 for a large range of reactions involving light particles ( nucleons , @xmath6 ) or photons on targets between the proton- and neutron - driplines more than 1200 cases are found with @xmath23 . although there is a dependence on spin ( in @xmath24 and selection rules for @xmath19 ) , a dependence on the coulomb barrier can be clearly seen : larger coulomb barries allow @xmath25 for larger @xmath26 , i.e. transitions with larger and larger energies are suppressed .
examples are shown in fig .
[ fig : qdep ] for ( p , n ) and ( @xmath6,n ) across the nuclear chart . there is a clear correlation between the highest @xmath26 appearing and the coulomb barrier ( defined by charge @xmath27 ) .
as expected the dependence on @xmath27 is stronger for reactions with the more highly charged @xmath6 than those involving protons .
the scatter for a given @xmath27 is rather determined by the available @xmath1 values than by spin effects .
this coulomb suppression of the stellar enhancement @xmath22 is important for the conception of experiments because it implies that it is sometimes better to measure in the direction of negative reaction @xmath1 value when being interested in results as close as possible to stellar values .
an example is the recently studied reaction @xmath28rb(p , n)@xmath28sr with @xmath29 mev @xcite .
9 c. freiburghaus , et al , astrophys .
j. 516 ( 1999 ) 381 .
h. schatz , et al , phys .
294 ( 1998 ) 167 .
m. arnould , s. goriely , phys . rep . 384 ( 2003 ) 1 .
t. rauscher , f .- k .
thielemann , at .
data nucl .
data tables 75 ( 2000 ) 1 .
t. rauscher , phys .
c , in press ( arxiv:0807.3556 ) .
t. rauscher , f .- k .
thielemann , k .-
kratz , phys . rev .
c 56 ( 1997 ) 1613 .
t. rauscher , et al , phys .
c 57 ( 1998 ) 2031
. g. g. kiss , t. rauscher , gy .
gyrky , a. simon , zs .
flp , e. somorjai , phys .
, in press .
g. g. kiss , et al , this volume . | nucleosynthesis processes involve reactions on several thousand nuclei , both close to and far off stability .
the preparation of reaction rates to be used in astrophysical investigations requires experimental and theoretical input . in this context
, two interesting aspects are discussed : ( i ) the relevant @xmath0 transition energies in astrophysical capture reactions , and ( ii ) the newly discovered coulomb suppression of the stellar enhancement factor .
the latter makes a number of reactions with negative @xmath1 value more favorable for experimental investigation than their inverse reactions , contrary to common belief .
address = departement physik , universitt basel , ch-4056 basel , switzerland |
it is well known that 3d spin systems on finite volumes show domain walls separating coexisting phases which behave as _ fluid interfaces _ between the critical and the roughening temperature .
the finite - size effects in the free energy of a fluid interface are dominated by long - wavelength fluctuations and a correct physical description of the critical properties of the surface can not neglect their contributions @xcite . while below the roughening temperature , where the interfaces are almost rigid , a microscopical approach can be taken ( see @xcite and references therein ) , above it one
is forced to assume an effective model describing the collective degrees of freedom of rough interfaces . the _ capillary wave model _ ( cwm )
@xcite in its simplest formulation , which we will follow , assumes an effective hamiltonian proportional to the area of the surface .
it has been recently shown @xcite that rather strong finite - size effects , depending on the shape of the lattice , are well described in terms of the _ one loop _ or gaussian approximation to the cwm : its predictions have been tested with high accuracy in the scaling region of the 3d ising model .
these one loop corrections depend only on one adimensional parameter , namely on the asymmetry @xmath0 of the transverse sizes of an elongated lattice @xmath1 ( @xmath2 ) with periodic boundary conditions taken in all directions @xcite .
as it has been already pointed out @xcite , higher order corrections to the gaussian model can be taken into account to verify the cwm beyond the one loop approximation .
these higher order corrections can provide a more stringent test of the cwm model : in fact , while many different effective hamiltonians reduce to the gaussian form at one loop level @xcite , they differ in the form of two and higher loop corrections . in this paper
we apply the cwm in the two loop approximation to the finite size behavior of order order rough interfaces in the 3d three - state potts model .
the two loop contributions do not depend only on the asymmetry parameter @xmath3 but also on an adimensional parameter proportional to the minimal area of the surface , namely @xmath4 , @xmath5 being the ( reduced ) interface tension .
an important consequence of this fact is that the well known finite - size behavior @xcite of the energy splitting @xmath6 occurring between vacua on finite volumes , has no longer , at two loop , the _ classical _ functional form @xmath7 not even for symmetric ( @xmath8 ) lattices .
this must be contrasted with what happens at one loop level , where , due to the scale invariance of the one loop contributions , one has to consider asymmetric lattices to find deviations from the classical functional form @xcite . to verify the cwm we study the finite - size behavior of rough order - order interfaces of the 3d three - state potts model by means of monte carlo numerical simulations .
this model is of great interest because of its well known connection with 4d @xmath9 pure yang
mills theory at finite temperature @xcite .
in fact one can assume an effective action of a 3d spin model with short range , ferromagnetic interaction to describe the finite temperature deconfinement transition of qcd in the limit of infinite quark masses @xcite .
for this reason the potts model has been extensively investigated and , in particular , the properties of order - order and order - disorder interfaces have been already studied @xcite , one of the main goals being the evaluation of the corresponding interface tensions . in this paper
we show that the knowledge of the functional form of the fluid interface free energy on finite volumes , including the capillary wave contributions , enables one to estimate the order - order interface tension with high precision .
moreover , the good agreement we find with the theoretical predictions of the cwm model at two loop order strongly supports the idea of the universality of this description of rough interfaces in 3d statistical models .
the paper is structured as follows : in sec .
we describe the cwm and its one loop and two loop approximations . in sec .
we compare the results of mc simulations with the cwm predictions .
is devoted to some concluding remarks .
according to the cwm , the interface between two domains of different magnetization in the ordered phase of a 3d spin system , above the roughening temperature , is described by the partition function @xmath10\exp\left\{-\sigma a\left [ x\right ] \right\}~~ , \label{zeta}\ ] ] where the single - valued function @xmath11 describes the displacement from the equilibrium position of the interface , @xmath5 is the reduced ( order - order ) interface tension and @xmath12 $ ] is the area of the interface @xmath13=\int_0^{r } dr \int_0^{t}dt~ \sqrt{1+\left(\frac{\partial x}{\partial r}\right)^2 + \left(\frac{\partial x}{\partial t}\right)^2}~~. \label{acw}\ ] ] it should be mentioned that ( [ acw ] ) coincides with the nambu - goto string action in @xmath14 in a particular gauge .
( [ acw ] ) is not expected to be the _ exact _ action describing fluid interfaces but at least the dominant contribution : as we will show this is indeed the case . to compare the predictions of the cwm with numerical results from monte carlo ( mc ) simulations , we have chosen 3d lattices of @xmath15 sites , with @xmath2 , and periodic boundary conditions in each direction .
this particular choice of the lattice shape allows one to consider only interfaces orthogonal to the elongated ( @xmath16 ) direction , the probability of having interfaces orthogonal to the other directions being negligible .
the 2d field @xmath11 is therefore defined on the rectangle @xmath17\times \left[0,t\right]$ ] with opposite edges identified , i.e. on a torus .
the partition function ( [ zeta ] ) can not be computed exactly , but it is possible to express it as an expansion in powers of the adimensional parameter @xmath18 : the two loop expansion of @xmath19 can then be written as @xmath20 the one loop contribution ( namely the gaussian approximation ) , obtained retaining only the quadratic term in the expansion of ( [ acw ] ) , is nothing else than the exact partition function of a 2d conformal invariant free boson on a torus of modular parameter @xmath21 @xcite @xmath22 the two functions @xmath24 and @xmath25 appearing above are respectively the dedekind function and the second eisenstein series : @xmath26 the three state potts model is defined by the partition function @xmath27\right\}\ ] ] where the variables @xmath28 are defined on a three dimensional hypercubic lattice and take the values @xmath29 in the termodynamic limit the potts model is known to undergo a ( weak ) first order phase transition at @xmath30 @xcite and the roughening temperature can be estimate to be @xmath31 @xcite . for @xmath32
the @xmath33 symmetry is spontaneously broken and the three ordered phases coexist , while at @xmath34 also the disordered phase coexists with the previous ones .
in the finite _ cylindric _ geometry we are considering spontaneous symmetry breaking at low temperature can not occur : the degeneracy of the ground state is removed , the energy of the symmetric , @xmath33 invariant , ground state being separated by an energy splitting @xmath6 from the two degenerate mixed - symmetry states .
the energy splitting is due to tunneling between the phases and is directly linked to the free energy of the interface @xcite .
according to the cwm , for @xmath35 , we assume @xcite , @xmath36 , @xmath37 where @xmath38 is an unpredicted constant and a convenient normalization has been chosen .
we would like to stress that the two loop contribution ( [ 2loop ] ) does not depend only on the ratio @xmath0 , like the one loop term ( [ 1loop ] ) , but also on the minimal area @xmath39 .
if we put @xmath40 , @xmath41 , in ( [ e1 ] ) we obtain @xmath42\right\}~~ , \label{e2}\ ] ] where @xmath43 .
the _ classical _ formula @xcite @xmath44 can be recovered only neglecting the two loop contribution .
the comparison between formula ( [ e1 ] ) and the values of @xmath6 extracted from mc simulations provides a simple and stringent way to verify the cwm predictions .
it should be noted that no new free parameters are introduced within this approach : the formulae ( [ e1 ] ) and ( [ e3 ] ) contain the same number of undetermined parameters , namely @xmath5 and @xmath38 .
to extract the energy splitting @xmath6 from mc generated ensembles we follow the procedure of @xcite . defining the time slice magnetization @xmath45 we compute the correlation function @xmath46 where @xmath47 , and we extract the transfer matrix low energy levels from the asymptotic @xmath48dependence of @xmath49 @xmath50\right\}\nonumber\\ & & + c_1\left\{\exp(-ke')+\exp[-(l - k)e']\right\}~+~\dots~~\label{fit}\\ z & = & 1 + 2 e^{-l e } + \dots\label{gk}\end{aligned}\ ] ] @xmath51 tr @xmath52 being the partition function ( the next to leading energy level @xmath53 turns out to be non negligible in our range of parameters ) .
having so extracted the energy splitting @xmath6 from the mc data for different values of the lattice sizes , we can compute the order - order interface tension @xmath5 and the constant @xmath38 by fitting our data with the formula ( [ e1 ] ) .
we have performed our simulations at @xmath54 , the longest lattice size being fixed at @xmath55 , and the other sizes varying in the range @xmath56 , @xmath57 , ( @xmath36 ) .
this value of @xmath58 is enough inside the ordered phase to make highly suppressed the probability of formation of order - disorder interfaces @xcite but presents a correlation length large enough to make the lattice artifacts negligible and to consider domain walls as fluid interfaces .
the fact that the disordered phase is substantially absent at this @xmath58 can be seen from the histograms of the real part of the magnetization @xcite @xmath59 as is shown , for example , in fig.1 .
the modulus of the magnetization at this @xmath58 is about @xmath60 for the lattice sizes we are considering : in this figure the projection on the real axis of the ordered phases are clearly visible while the peak centered at @xmath61 , which would signal the presence of the disordered phase , is absent . fig .
2 represents a typical distribution of the magnetization @xmath62 for a sample of our mc generated configurations .
most configurations consist of a single phase or of two phases separated by two interfaces ( the minimum number compatible with periodic boundary conditions ) .
the single phase configurations are represented by the three clusters of points corresponding to the three degenerate vacua ; the two interface configurations form the straight lines joining these clusters .
interface configurations , which tend to fill uniformly the interior of the triangle , are clearly visible in fig .
3 : it corresponds to a @xmath63 , @xmath64 lattice and to a larger probability of having tunneling events ( i.e. interfaces ) , while these are much more suppressed for the lattice of fig .
2 ( @xmath65 ) .
we have used a swendsen - wang cluster algorithm @xcite to perform our mc simulations . to keep under control correlations in mc time and cross - correlations between the @xmath49 observables
, we have systematically scattered our measurements avoiding the measurement of two different observables at the same mc time .
we have made between @xmath66 and @xmath67 sweeps for each experiment , depending on the lattice size , obtaining about @xmath68 data per observable .
however , the covariance matrix turns out to be different from the diagonal form , which one expects from a sample of statistical independent data .
we have taken this fact into account by including the covariance matrix in the fitting procedure to formula ( [ fit ] ) to extract the energy gap @xmath6 : the results are reported in tab .
the error on @xmath6 has been estimated with an ordinary jackknife procedure .
0.3 cm [ cols="^,^,^,^,^,^",options="header " , ] * tab .
_ results of the fit of @xmath6 with two loop , one loop and classical approximations of the cwm , considering all values of @xmath69 , symmetric @xmath70 lattices or ( r , t=11 ) lattices . _ 0.3 cm on the other hand , the result obtained using the classical formula ( [ class ] ) is not compatible with the previous ones even using only the symmetric data , as it is shown in the fifth line of the same table . the best fit curve obtained from ( [ class ] ) in the latter case
is plotted in fig .
5 were also the `` asymmetric '' mc data are reported for comparison .
we would also like to observe that a good agreement with the one loop approximation of the cwm can be obtained @xcite if one considers low values of ratios @xmath0 and high values of the minimal area @xmath39 , i.e. where the two loop contribution ( [ 2loop ] ) are maximally suppressed .
this is show in the last two lines of tab . 2 and in fig.6 .
in this paper we have shown that the cwm in the two
loop approximation provides an excellent description of order order interfaces in the 3d three state potts model .
this result , together with the corresponding one for the 3d ising model @xcite , strongly supports the hypothesis of the universality of the cwm description of interface physics in 3d statistical models .
it is worth stressing again that the cwm corrections ( [ e1 ] ) to the finite size behavior of the interface free energy do not introduce any new free parameters with respect to the `` classical '' picture ( [ class ] ) . besides the intrinsic physical interest of this picture
, it should be noted that the cwm provides an accurate description of finite size corrections to the free energy of rough interfaces , thus enabling one to extract correct informations about physical observables from finite lattices of different geometries , as it has been recently shown @xcite .
0.5 cm * acknowledgements * 0.5 cm we are grateful to m. caselle and f. gliozzi for many useful suggestions
. one of us ( s.v . ) would like to thank s. alves dias , m. g. do amaral , c. a. linhares , j. mignaco and m. a. do rego monteiro for stimulating discussions and the cbpf for the kind and warm hospitality in rio de janeiro .
the work of p.p . is supported by an infn grant . the work of s.v . in rio
de janeiro is supported by an cnpq grant .
0.5 cm * fig . 1 . * _ histogram of the real part of the magnetization for a typical mc ensemble ( in this case @xmath71 , @xmath64 ) .
the absence of a peak in @xmath72 indicates that the disordered phase does not coexist with the ordered ones at our @xmath54 . _
0.5 cm * fig .
2 . * _ distribution of a sample of 5,000 configurations generated by the mc simulation in the complex plane of the magnetization for a lattice @xmath65 .
the three clusters of points represent the one phase configurations ; the straight lines joining the clusters are the two interface configurations . _ 0.5 cm * fig .
3 . * _ the same of fig . 2 for a lattice @xmath63 , @xmath64 .
the three - interface configurations are uniformly distributed in the interior of the triangle . _
0.5 cm * fig .
4 . * _ comparison of the predictions of the cwm with the mc data : the energy gap @xmath6 is plotted as a function of @xmath0 .
the lines represent the best fit of all data to formula ( [ e1 ] ) : from up to down they correspond respectively to @xmath73 fixed ( cfr . tab .
1 ) . _ 0.5 cm * fig . 5 . * _ comparison of the predictions of the classical formula with the mc data : @xmath6 is plotted as a function of @xmath39 .
the line represent the best fit of the symmetric @xmath74 data , reported with error bars , to formula ( [ class ] ) ; the asymmetric mc data are also reported : squares correspond to @xmath75 data , circles to @xmath63 and diamonds to @xmath76 . _
0.5 cm * fig . 6 . *
_ comparison of the predictions of the two loop ( [ e1 ] ) and one loop ( [ 1lupo ] ) approximations for the @xmath76 data , with @xmath0 . the dashed line represent the one loop best fit while the two loop is the dotted line . _ 0.3 cm | the physics of fluid interfaces between domains of different magnetization in the ordered phase of the 3d three state potts model is studied by means of a monte carlo simulation .
it is shown that finite size effects in the interface free energy are well described by the capillary wave model at two loop order , supporting the idea of the universality of this description of fluid interfaces in 3d statistical models .
theoretical physics institute + university of minnesota 0.5 cm tpi
minn93/46t + cbpf nf058/93 + dftt 58/93 + october 1993 0.5 cm * capillary wave approach to order - order fluid * 0.2 cm * interfaces in the 3d three - state potts model * paolo provero + _ theoretical physics institute _ + _ university of minnesota _ + _ minneapolis , mn 55455 , usa _ 0.2 cm stefano vinti + _ cbpf _ + _ centro brasileiro de pesquisas fisicas _ + _ rua dr .
xavier sigaud 150 _ + _ 22290 rio de janeiro , brasil _ + _ and _ + _ dipartimento di fisica teorica _ + _ universit di torino _ + _ via p. giuria 1 _ + _ 10125 torino , italia _ 0.5 cm |
numerous experiments have exhibited the probabilistic nature of our universe at the quantum scales .
a further insight into this probabilistic character was developed by richard feynman with the path integral formulation of quantum mechanics @xcite . in this formulation
the transition of a particle between two points is described by taking into account all possible particle trajectories between these points .
each path contributes a phase @xmath0 , where @xmath1 is the action for that path .
the amplitude or the propagator is the sum of the contributions from all the trajectories . inspired by the path integral formalism ,
we describe here a comparatively simplified approach that leads to the euclidean propagator of a free particle .
in addition , this model can provide further insights into motion of the particle at quantum scales .
these insights include the limit on the step size for the particle s motion .
we will show that the euclidean propagator in this scenario is valid for step sizes of the order or greater than the de broglie wavelength of the particle .
we also describe the statistical ensembles that characterize this transition .
a motivation to introduce these ensembles is the relationship between the time evolution operator in quantum mechanics with the partition function as @xmath2 , through the transformation @xmath3 .
the approach considered in this article overlaps with previous studies on the statistical approach to quantum mechanics . in reference
@xcite , for instance , monte carlo methods were used to evaluate the euclidean path integral with individual paths of the particle defined on a discrete time lattice .
these methods were then used to treat the anharmonic oscillator in 1 spatial dimension .
here , we shall focus on the motion of a free particle and derive the propagator using numerical techniques with mathematica @xcite . moreover
, we shall derive a limit on the minimum step size of the particle and extend our approach to more than one dimensions in the section [ model-2d ] .
the paper is organized as follows : in section [ model-1d ] we describe the model we employ in one dimension .
we show that the sum over all paths leads to the euclidean propagator and also present a possible definition of the probability for each path .
we then describe the statistical mechanics model that mimics this transition .
section [ model-2d ] generalizes the model to two and higher dimensions .
we conclude in section [ conclude ] . .
consider the one dimensional motion of a free particle between two points a and b in discrete spatial steps of size @xmath4 .
the particle covers a distance @xmath5 in time @xmath6 . by discrete steps
we assume that the particle ` vanishes ' at one point and appears at the other after time @xmath6 .
we will describe in subsection [ action-1d ] what we mean by vanishing .
our aim is to choose @xmath4 as small as possible in order to describe the continuous motion of the particle in good approximation .
the particle starts at a and reaches point b after a minimum number of @xmath7 steps , i.e. @xmath8 .
the particle can move from point a to b by taking all forward steps or can flip @xmath9 times as it moves towards b. each subsequent path contains an additional flip .
therefore the number of steps in subsequent paths are given by @xmath10 where the @xmath9th trajectory has @xmath11 steps .
the number of possible trajectories the particle can take and the entropy for the @xmath9th path is given by @xmath12 where @xmath13 represents the number of forward steps and @xmath14 represent the number of backward steps in the @xmath9th path . note
also that @xmath15 and @xmath16 so that the net displacement of particle is @xmath17 and @xmath18 is the total distance traveled . for example , with @xmath19 and @xmath20 , @xmath21 implies that there are 4 possible trajectories the particle can take in which it flips once in each .
these four paths are shown in fig .
the arrows in the figure show the direction of motion and not the actual path of the particle .
the number of possible paths @xmath22 increase as @xmath9 increases and so does the entropy @xmath23 .
however the ratio of entropy and the number of steps approaches a constant value @xmath24 as @xmath25 .
the entropy is zero for the classical path . ) as a function of the dimensionless parameter @xmath26 for different values of @xmath27 .
the thick colored lines show the numerical calculation of the sum over paths in ( [ prop-1d ] ) using mathematica whereas the dashed colored lines shows the decaying exponential function .
we can see that the numerical sum approaches the exponential function exp@xmath28 for @xmath29 for each colored line .
this implies that the numerical sum approaches the euclidean propagator for @xmath30 .
[ fig - propagator ] ] we assume that the particle moves with uniform velocity @xmath31 in each of the possible paths . for the @xmath9th path the particle travels a distance @xmath18 in time @xmath6 .
we define the speed of the particle for the @xmath9th path as follows : @xmath32 we further define the action for the @xmath9th path as @xmath33 where @xmath34 and @xmath35 is the mass of the particle . the amplitude is given by summing the contributions of all the paths between the two points , @xmath36 where @xmath37 is the reduced planck s constant and @xmath38 is a dimensionless constant . we can calculate the sum in equation ( [ prop-1d ] ) numerically using mathematica @xcite and fig [ fig - propagator ] shows the plot of this function for different values of @xmath7 .
we can see that for @xmath39 the sum approaches the limit @xmath40 .
the condition @xmath39 translates into the requirement @xmath41 , where @xmath42 is the de broglie wavelength of the particle .
we therefore obtain @xmath43 only if @xmath44 where , @xmath45 and @xmath46 .
we choose @xmath47 to get the correct normalization of the propagator .
therefore , for a particle initially at @xmath48 the propagator is given by @xmath49 which is the solution of the heat equation .
wick rotating time @xmath50 yields the free particle propagator which is the solution of the schrodinger s equation .
note that the sum in equation ( [ prop-1d ] ) approaches infinity for @xmath51 and the above propagator is not valid for @xmath52 .
the allowed range for @xmath4 is therefore @xmath53 .
earlier we assumed that the particle vanishes at one point and appears at the next step .
this limit on @xmath4 sheds light on what we meant by ` vanishing ' .
it means that at quantum scales this model describes the behavior of particles for @xmath54 .
in other words we assume that at the quantum scale a more fundamental formalism describes the particle s motion for lengths less than the de broglie wavelength of a particle . ) as a function of the number of times the particle can flip in a particular path . for a fixed step value @xmath8 the probability decreases as @xmath9 increases . as the number of steps @xmath7 increases and the trajectory becomes more continuous the probability of the particle to follow the classical path increases . from equation ( [ limit-1d ] ) , the classical limit
@xmath55 implies the continuous limit @xmath56 . for this limit
the probability for the classical path is 1 and for all other paths is zero . for quantum scales with @xmath57
, the above plot shows that the probability of the particle to choose other paths is non - zero but small .
moreover , with @xmath58 the quantum effects are more prominent for smaller distances .
[ fig - prob ] ] one possible way of parameterizing the probability of the particle to choose the @xmath9th path is as follows : @xmath59 where the denominator normalizes the sum of the probability to 1 .
note also that @xmath60 .
a drawback of defining the probability as in equation ( [ prob-1d ] ) is that the probabilities for paths other than the classical paths are too small .
note also that equation ( [ prob-1d ] ) is not a unique way to parametrize the probabilities .
another possible way is given in equation ( [ prob-1d - b ] ) of the appendix b. fig .
( [ fig - prob ] ) shows the plot of the probability as a function of @xmath9 for different values of @xmath7 .
we can see from fig .
( [ fig - prob ] ) that for large values of @xmath7 the probability for the classical path approaches 1 .
this is because for large values of @xmath7 the path next to the classical path has a large multiplicity and therefore a small probability .
in fact , for @xmath56 , the probability @xmath61 and @xmath62 .
in other words , as @xmath56 ( @xmath51 ) the path next to the classical path has so much entropy that the particle does not traverse it .
this implies that the limit @xmath56 can be chosen as the classical limit .
we described earlier that we need @xmath4 to be as small as possible in order to describe the continuous motion of the particle in good approximation .
as shown in equation ( [ limit-1d ] ) , the euclidean propagator is obtained if the lower bound on @xmath63 .
this is of course true for quantum scales but in order to describe classical objects we can choose @xmath51 which is the limit @xmath55 . if we choose @xmath63 the probability for paths other than the classical path is small but non - zero .
the fluctuations in the particle s trajectory get stronger for small distances .
the average distance , mean squared distance and the standard deviation for the @xmath9th path traveled by the particle , i.e. , @xmath64 where @xmath4 is the step size . in the above equation @xmath65
is the displacement of the particle and @xmath66 is the total distance traveled by the particle in the @xmath9th trajectory .
@xmath67 measures the deviation from the mean path and the paths with large deviations are less probable since the probability decreases sharply with @xmath9 according to our choice of the probability in equation ( [ prob-1d ] ) , also shown in fig [ fig - prob ] . this transition can be modeled on ensembles of 2 level spin systems .
each of our spin system is a canonical ensemble which contains @xmath68 number of localized spin 1/2 particles which are placed in a magnetic field .
there are two possible energy states of each particle , @xmath69 and @xmath70 .
we have chosen a two level system because a spin up ( @xmath69 ) represents a step forward on @xmath71-axis and a spin down ( @xmath70 ) represents a step backward .
all our ensembles will have the same net magnetic moment since this represents the displacement of the particle .
the partition function of the @xmath9th ensemble is given by @xmath72 where @xmath73 and @xmath74 is the boltzmann constant .
the requirement that every subsequent ensemble has one additional spin up or down leads to the following parametrization of the probability and the number of particles @xmath75 here , @xmath76 ( @xmath77 ) is the probability of spin up ( spin down ) .
similarly , @xmath78 ( @xmath79 ) is the number of spin up ( spin down ) particles in the @xmath9th ensemble .
for a two level system with @xmath68 number of spins the average energy , mean square energy and the fluctuation in energy are given by @xmath80 these are similar to the expressions for distance of the particle in equations ( [ distance-1d ] ) . in the standard quantum mechanical picture the classical limit is recovered as @xmath55 . in our set of ensembles
this limit corresponds to @xmath81 . in this case
the partition function would be @xmath82 and this system represents the path with only forward steps which is the classical path . in our ensemble therefore @xmath55 is equivalent to @xmath81 .
in this section we extend our model in 1-d to two and higher dimensions . in two dimensions
the method is somewhat different from the one we adopted in section [ model-1d ] .
we start with a 2d model which generalizes the ideas described in the previous section .
we will first see that in order to represent the particle s motion we need to rotate the frame so that the particle s classical trajectory is on one of the axis . in this case
the particle s motion will be along one direction with the fluctuations being in the perpendicular direction .
let us first assume that the particle moves from the origin to a point @xmath83 in the @xmath84 plane .
the multiplicity and entropy in this case are @xmath85 @xmath86 where @xmath87 is the total number of steps the particle takes in the @xmath71 and @xmath88 directions .
@xmath89 quantifies the possible paths a particle can take .
for instance , with @xmath90 and @xmath91 , we get @xmath92 .
these 6 paths are shown in fig [ fig - paths22 ] .
these paths are the minimum distance paths the particle can take .
so , in 2 dimensions there are several routes that corresponds to the minimum distance of travel for the particle .
we can see this from equation ( [ w-2d ] ) .
for @xmath90 we get @xmath93 therefore , there are several paths that corresponds to the minimum distance path .
this is due to the fact that the particle can only take discrete steps . in more than one dimension
therefore we will rotate our coordinate system so that the particle lies on the one of the axis ( say @xmath94-axis ) of the new coordinate system .
on rotating our coordinate system the new distance @xmath94 of the particle from the origin will be our euclidean distance in the original frame ( @xmath95 ) @xcite .
this implies that @xmath96 in our rotated coordinate system . in the rotated system ,
the distance traveled by the particle in path ( @xmath97 ) with @xmath68 steps in the @xmath94 direction and @xmath98 steps in the @xmath99 direction is @xmath100 . in this coordinate system
the particle starts at @xmath101 .
we therefore place @xmath96 in equation ( [ w-2d ] ) and get @xmath102 @xmath89 now quantifies the number of paths in the rotated frame . fig .
[ fig - pathk1 ] shows the possible paths for the particle when it flips once in the @xmath99-direction ( @xmath103 ) . for the paths with @xmath103 , @xmath104 and @xmath105
there are 12 equally probable trajectories for the particle .
we can parametrize the probability for this case as before : @xmath106 the probability in this case also behaves similar to the 1 dimensional case .
the difference is that sum now is over all the flips in the @xmath71 and @xmath88 directions .
for example , for @xmath107 , the probability of the path which include one flip in the @xmath71 and @xmath88 directions is @xmath108 . in this case
the probability will be suppressed more since the number of possible paths have now increased . however , in the limit of @xmath109 the probability for all paths except the classical path is zero . ) for the particle to take for @xmath90 and @xmath91 ( @xmath92 ) .
all the paths shown are the minimum distance paths when the particle takes discrete steps .
[ fig - paths22 ] ] next , following steps similar to section [ model-1d ] we evaluate the propagator of the particle in two dimensions .
the distance traveled by the particle in any step is @xmath100 , where the @xmath110 corresponds to the fluctuation in the @xmath99 direction .
the speed of the particle for the path ( @xmath97 ) is : @xmath111 as before , we define the action as @xmath112 where we have assumed @xmath113 .
the amplitude in this case is given by @xmath114 where @xmath38 .
we can again calculate this sum numerically and obtain a similar limit on @xmath4 as before .
therefore , @xmath115 which is true if @xmath116 where we have placed @xmath117 to obtain the propagator in our original frame .
we can use this strategy in any number of spatial dimensions .
the constant @xmath118 can be chosen as in section [ action-1d ] .
-direction , i.e. , @xmath119 and @xmath105 ( @xmath120 ) .
fig 5(a ) shows 6 paths together where the particle can flip ( double red lines ) in the @xmath88 direction at any of the points in the upward or downward direction .
fig 5(b ) shows the other 6 paths the particle can take for this case .
[ fig - pathk1 ] ] in this case the ensemble is a system with two different types of particles , namely type @xmath121 and @xmath122 .
we assume that a mechanism maintains the two spins at different temperature , @xmath123 and @xmath124 .
in addition , a constraint is imposed on the system so that the number of spin up or down particles of a particular type are independent of the other .
this is done so that this system represents the motion in two independent directions @xmath71 and @xmath88 . at each location in the ensemble
there can be either particle of type 1 or 2 with spin up or down .
the partition function of the system is given by @xmath125 where @xmath126 and @xmath127 are energies of the two types of particles .
we further impose the following restriction on the system described by the partition function in equation ( [ pf-2d ] ) @xmath128 this condition renders the distribution of spins independent of each other .
again , this is to take into account the fact the motion of the particle along one axis is independent of the motion along the other . using the above condition ,
the partition function in ( [ pf-2d ] ) for @xmath129 particles can be written as @xmath130 the average energy , mean square energy and the fluctuation in energy for each of the two spin types are given by equations similar to equations ( [ energy-1d ] ) .
the entropy of this system is given in equation ( [ entropy-2d ] ) of the appendix and agrees with that obtained from ( [ s-2d ] ) in the large @xmath131 limit .
we next require that there are no spins of type 2 in the minimum entropy ensemble , i.e. , @xmath96 .
this amounts to rotating the coordinate system .
with @xmath96 the probability of the spin up or down is equal to 1/2 for the spin of type 2 .
equation ( [ pf-2d - b ] ) becomes @xmath132 the above partition function represents the set of ensembles that model the transition of the particle in 2 dimensions with the rotated frame .
an approach similar to that described in previous sections can be used to describe the particle s transition in any number of dimensions .
the number of steps in 3d will be @xmath133 .
we can similarly rotate the coordinate system so that the particle lies on one of the axis ( say @xmath94 axis , @xmath134 ) .
the multiplicity in this case will be @xmath135 where @xmath136 now is the number of steps in the @xmath137 direction .
the probability can be defined as in equation ( [ prob-2d ] ) .
the statistical ensemble would have a third type of spin and a condition similar to ( [ rest-2d ] ) would render the number of spins independent of each other . therefore , the analysis presented in these section can be used to extend the model to any number of spatial dimensions .
we considered a simplified model of the transition of a particle between two points in discrete spatial steps .
the aim was to study how the sum over histories or the path integral interpretation of quantum mechanics can be implemented in this scenario .
this model is presented as an approximate description of the standard approach which involves integrating over all possible trajectories of the particle to derive the free particle propagator .
we started with the particle s transition in one dimension and included trajectories in which the particle can flip any number of times . for each path we defined a multiplicity and thereby an entropy .
we showed that the sum over all paths in this case leads to the euclidean propagator for the particle if the step size is taken to be less than or of the order de broglie s wavelength ( @xmath42 ) .
the propagator we obtained was valid for step size @xmath54 . in the classical limit ( @xmath55 )
the number of steps approach infinity and in this limit the classical path is the only contributing path . the probability of each path was taken to be proportional to the inverse of multiplicity .
the paths involving large flips or fluctuations were shown to be less probable according to our definition .
the classical path was defined to be the trajectory for which the number of steps @xmath56 ( @xmath51 ) . as @xmath56 the path next to the classical path has so much entropy
that the particle does not traverse it .
we then extended this description to two dimensions .
since there are several possible minimum distance paths for the particle in 2d we showed that this scenario can be described by rotating the frame such that the two points are on one of the axis of the rotated frame .
the free particle propagator can be derived for this case also leading to the same limit on the spatial step .
we interpreted this limit on the spatial step as the limiting length to which this model is valid and below which a more fundamental formalism describes the particle s motion .
we also described the statistical ensembles that correspond to these transition . in 1d
the ensemble is a two level system placed in a magnetic field . in 2d
the ensemble is also a two level system but with two different types of spins .
additional constraint was imposed on this system so that the number of spins at any given temperature is independent of the other .
this was to take into account the fact that the motion of the particle along one axis is independent of the motion along the other .
for these ensembles the limit @xmath81 corresponds to the classical limit of the particle s transition .
the ideas in 2 dimensions can be used to extend this model to any number of spatial dimensions .
we have therefore demonstrated the possible implications of the feynman path integral formulation of quantum mechanics in a model which describes the transition of a particle between two points in discrete spatial steps .
the author would like to thank fariha nasir and muhammad shahbaz for useful discussions .
here we present the expression for the entropy of the system of ensembles in 1 and 2 dimensions . for 1d , in the large @xmath131 limit , the entropy of a two level system described by the partition function in ( [ pf-1d ] ) is given by @xmath138- \beta_j e \tanh(\beta_j e ) \right ) \label{s-1d - a}\end{aligned}\ ] ] the condition that subsequent ensembles include one additional spin up and down translates to the following condition @xmath139 which implies @xmath140 using above equations we can write the entropy of the two level in equation ( [ s-1d ] ) system as @xmath141 \label{s-1d - b}\end{aligned}\ ] ] similarly , for the system with two different types of spin 1/2 particles , the entropy is given by @xmath142 where @xmath143 and @xmath144 are the entropy of spins of type 1 and 2 given by expressions similar to ( [ s-1d - b ] ) .
note that in the above expression 1 and 2 represents the type of spin and we have suppressed the subscript 1d .
the probability of the particle to choose the @xmath9th path can also be quantified as follows : @xmath145 where @xmath146 is the number of possible ways a particular path can be traversed and is given in equation ( [ w-1d ] ) . also , @xmath76 is the probability of a step forward and @xmath77 is the probability of a step backward .
the other parameters can be parameterized as in equations ( [ ens-1d ] ) @xmath147 the probability chosen in this manner decreases slowly with @xmath9 compared to equation ( [ prob-1d ] ) .
a problem with this choice however is that the sum @xmath148 of the probability is not convergent in 1d . | we study the transition of a particle between two points such that the particle takes discrete spatial steps in this transition .
we analyze how the sum over histories interpretation of quantum mechanics can be implemented in this scenario .
we show that the euclidean propagator of a free particle is recovered if the minimum space interval is of the order or greater than the de broglie s wavelength of the particle .
we also describe the statistical ensembles that model this transition .
furthermore , we discuss a possible extension of this model to 2-dimensions which serves as an example to extend it to any number of spatial dimensions .
* sum over histories : discrete step interpretation * muhammad adeel ajaib _ university of delaware , newark , de 19716 , usa _ |
the sedimentation of a heavy solid in a fluid has been studied thoroughly , as the dynamics of settling and sliding are relevant to a broad class of phenomena across many orders of magnitude , ranging from landslides @xcite , earthquakes @xcite , avalanches @xcite , to the lubrication of cartilaginous joints @xcite , and motion of cells in a microfluidic channel @xcite or in a blood vessel @xcite . following the now classical studies of the dynamics of a particle near a rigid wall @xcite
, additional effects such as the influence of the boundary conditions @xcite , and their role on drag @xcite , viscometry @xcite , and bouncing @xcite , have been accounted for .
recently , motion of wedge - like objects down an incline @xcite , as well as the effects of elasticity in such contexts as granular impact @xcite , polymer - bearing contacts @xcite , solvent permeation in gels @xcite , soft lubrication @xcite , transient effects in displacement - controlled systems @xcite , settling on soft and poroelastic beds @xcite , adhesive walls @xcite , and self - similar contact @xcite , have also been addressed . in all these phenomena ,
the minimal model of motion relates to that of a solid object immersed in a viscous fluid in the vicinity of a soft elastic or poroelastic wall . perhaps surprisingly then
, the general theory for the free motion of a rigid solid close to a soft incline which has through its degrees of freedom the ability to simultaneously sediment , slide , and spin does not seem to have been considered . these modes naturally arise in several applications such as particle capture , joint lubrication , and have analogues in certain geophysical phenomena . here , we study this problem in a minimal setting and describe the essential scalings and qualitative features , develop a soft lubrication theory that complements these scaling ideas , and solve the resulting equations numerically to characterize the broad range of possible behaviours .
, inside a viscous fluid ( blue ) , in the vicinity of a thin soft wall ( brown ) .
the ensemble lies atop a tilted , infinitely rigid support ( grey).__,width=340 ] we consider the 2d system depicted in fig . [ fig1 ] that consists of the free gravitational fall of a long cylinder of radius @xmath0 , density @xmath1 , mass ( per unit length ) @xmath2 , and buoyant mass @xmath3 , where @xmath4 is the density of the neighbouring fluid of viscosity @xmath5 .
we assume that the motion of the cylinder occurs in the vicinity of a wall that is inclined at an angle @xmath6 $ ] with respect to the horizontal direction , and coated with a soft elastic layer of thickness @xmath7 , and lam s coefficients @xmath8 and @xmath9 .
we denote by @xmath10 the deformation of the fluid - wall interface .
note that a positive indentation of the compressible elastic wall corresponds to a negative value of @xmath11 .
the system is assumed to be invariant along @xmath12 , i.e. we limit ourselves to planar motions wherein the cylinder has three degrees of freedom : the gap @xmath13 between the cylinder and the undeformed wall along @xmath14 , the tangential coordinate @xmath15 of the cylinder axis along @xmath16 , and the angle @xmath17 through which the cylinder has rotated .
we further assume that the cylinder starts its motion at time @xmath18 , with @xmath19 , and possibly non - zero initial translational and angular velocity . due to this scale separation , we are in the lubrication regime @xcite , where the fluid viscous shear stresses are small relative to the flow - induced pressure @xmath20 contains only the flow contribution . ]
, which itself vanishes far from the contact zone as @xmath21 .
the tangential extent @xmath22 of the flow - induced pressure disturbance scales as @xmath23 , so that as for the hertzian contact @xcite , we can assume a parabolic shape of the deformed interface , and the total gap profile may be written as : @xmath24 ^ 2}{2r}\ .\ ] ] the thin soft compressible wall may also be treated via a lubrication - like theory for elastic deformations if is too small , and the condition @xmath25 is not satisfied . nevertheless , if for instance @xmath26 , this is a valid assumption since @xmath27 . ]
@xmath28 , so that the algebraic displacement @xmath10 of the fluid - wall interface is simply obtained from linear elastic response to the _ local _ flow - induced pressure disturbance @xcite : @xmath29 to characterize the motion of the cylinder near the inclined thin soft wall , we need to calculate the fluid drag force created by the flow - induced pressure field in the contact zone , which is driven by the tangential fluid velocity @xmath30 along @xmath16 .
we non - dimensionalize the problem using the following choices : @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , and @xmath39 , where we have introduced a free fall velocity scale @xmath40 , and the dimensionless parameter : @xmath41 this parameter measures the ratio of the free fall time @xmath42 and the typical lubrication damping time @xmath43 over which the inertia of the cylinder vanishes .
in fact , for a cylinder falling towards a rigid wall , the lubrication drag force ( per unit length ) exerted in the contact zone reads @xmath44 @xcite .
the typical decay time of the cylinder inertia can thus be estimated by balancing this damping force and the inertia ( per unit length ) @xmath45 , which leads to the above time scale .
note that the power @xmath46 is specific to the 2d case . with these definitions , the dimensionless gap profile given by eq .
( [ gapreal ] ) becomes : @xmath47 ^ 2+\kappa p(x , t)\ , \ ] ] where the dimensionless compliance is : @xmath48 to remain in the linearly elastic regime for the wall deformation , we assume that @xmath49 . before delving into a detailed theory , we first derive some scaling relations for the sliding dynamics of the cylinder . for steady motions , the analysis of @xcite shows that one non - trivial effect of the soft substrate is to induce a positive elastohydrodynamic pressure @xmath50 in the contact zone , when the particle is translated uniformly along the wall at speed @xmath51 while being at a constant distance @xmath52 .
in the present 2d - like case of a free cylinder , when that pressure is integrated once along the contact length @xmath53 , this leads to a net positive elastohydrodynamic lift force ( per unit length ) @xmath54 that tends to repel the sliding particle away from the soft wall . since the force of gravity ( per unit length )
@xmath55 tends to bring it back towards the wall , balancing the two forces allows one to predict a sliding height as a function of the speed @xmath51 , given by : @xmath56 when the sliding velocity @xmath51 does not vary much as is often the case on short time scales when damping in the normal direction @xmath14 is much stronger than in the tangential direction @xmath16 this represents a stable equilibrium gap thickness . a small perturbation about this equilibrium position suggests that the the cylinder will oscillate with frequency @xmath57 , even as the lubrication viscous damping will cause these inertial oscillations to decay over a typical time @xmath58 , as already introduced above . finally , after a transient evolution along the tilted wall
, we expect the cylinder to reach a long - term steady - state sliding regime characterized by a terminal velocity @xmath59 and a constant gap thickness @xmath60 .
leaving aside the conditions of existence of this scenario for now , we can already describe the properties of this regime using simple arguments . along @xmath14
, the gravity - vs - lift force balance leads to eq .
( [ deltaeq ] ) above , with @xmath61 and @xmath62 .
the second equation we need comes from the power balance in the direction of sliding motion @xmath16 .
the power ( per unit length ) @xmath63 generated by the gravitational driving is entirely dissipated in the contact zone through the viscous damping power @xmath64 .
this leads to the expressions for the steady gap and terminal velocity , given by : @xmath65 with these scaling relationships in place , we now aim at constructing a detailed soft lubrication theory that goes beyond these arguments and , as we will see , introduces new phenomena as well .
in the thin - gap limit , the governing stokes equations for incompressible viscous flow are given in scaled form by @xcite : @xmath66 together with no - slip boundary conditions , @xmath67 and @xmath68 .
solving eq .
( [ stokes ] ) with the above boundary conditions , and invoking the condition of volume conservation : @xmath69 yields the following equation for the evolution of the gap : @xmath70_x\ .\ ] ] ) , for @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , and @xmath77.__,width=510 ] the solution of this equation allows us to evaluate the total pressure - induced drag exerted on the cylinder , through : @xmath78 where we have used the fact that the normal vector to the cylinder surface is @xmath79 .
similarly , the shear drag force exerted on the cylinder along the @xmath80 axis is given by : @xmath81 when the dimensionless compliance is assumed to satisfy @xmath49 , we may employ perturbation theory in this parameter @xcite , using the following expansion for the pressure , @xmath82 , where @xmath83 .
as detailed in appendix a and b , integrating eq .
( [ eqgen3 ] ) to first order in @xmath84 , and using eqs .
( [ dragp ] ) and ( [ drags ] ) , leads to the following expressions for the perpendicular drag along @xmath85 and the two parallel components along @xmath80 : @xmath86\ , \\
d_{p,\parallel}&=&\pi\sqrt{2\epsilon}\ \frac{\dot{\theta}-\dot{x}_{\textrm{g}}}{\sqrt{\delta}}+\kappa\sqrt{\frac{\epsilon}{2}}\left [ \frac{23\pi\dot{\delta}\left(\dot{\theta}-\dot{x}_{\textrm{g}}\right)}{8\delta^{7/2}}+\frac{\pi\left(\ddot{x}_{\textrm{g}}-\ddot{\theta}\right)}{2\delta^{5/2}}\right]\ , \nonumber\\ d_{\sigma,\parallel}&=&-\pi\sqrt{2\epsilon}\frac{\dot{\theta}}{\sqrt{\delta}}+\kappa\sqrt{\frac{\epsilon}{2}}\left[\frac{\pi\left(\ddot{\theta}-\ddot{x}_{\textrm{g}}\right)}{4\delta^{5/2}}+\frac{\pi\dot{\delta}\dot{x}_{\textrm{g}}}{2\delta^{7/2}}-\frac{19\pi\dot{\delta}\dot{\theta}}{8\delta^{7/2}}\right]\nonumber\ .\end{aligned}\ ] ] we stress that we have neglected the forces acting outside the contact zone , consistent with the lubrication approximation .
to justify this choice , let us first consider the sedimentation motion towards the rigid wall .
the drag force ( per unit length ) exerted on a cylinder in a bulk fluid scales as @xmath87 @xcite . according to eq .
( [ 3drag ] ) , the pressure - induced lubrication drag force ( per unit length ) reads , in real variables , @xmath88 @xcite . since @xmath89 in the lubrication approximation , one can safely neglect the bulk drag against the lubrication one acting in the contact zone .
similarly , according to eq .
( [ 3drag ] ) , for the tangential motion along a rigid wall , the pressure - induced drag scales as @xmath90 @xcite , which despite being smaller than @xmath91 is once again larger than @xmath92 , in the lubrication approximation .
one can thus safely neglect the bulk drag relative to the lubrication drag for the tangential degree of freedom as well .
since the shear - induced drag is of the same order and symmetry as the tangential pressure - induced drag , the previous conclusion extends immediately to the rotational degree of freedom .
we note that the argument above assumes a rigid wall , since the fluid lubrication order is not modified by the softness of the wall . as an illustration of this statement , all the softness - induced terms the ones proportional to the independent compliance parameter @xmath84 in eq .
( [ 3drag ] ) have the same order in @xmath93 as the corresponding terms for the rigid wall
. we also note that it may not be satisfactory at first sight to get an acceleration - dependent drag , even as a first order correction , as it means at time @xmath94 , when there is no flow , there is a pressure field that deforms the wall . to understand this , we note that the origin of this behaviour is to be found in the @xmath95 term in eq .
( [ eqgen3 ] ) , since @xmath96 due to the stokes equation . in our analysis , we have neglected the linearized inertia of the fluid , @xmath97 , but at very short times this term becomes dominant and resolves this apparent paradox . ) , for @xmath98 , @xmath99 , @xmath73 , @xmath74 , @xmath75 , @xmath100 , and @xmath101 .
if we replace the last condition by @xmath102 , then @xmath103 diminishes.__,width=510 ] knowing the dominant elastohydrodynamic drag forces acting on the cylinder , we now use the balance of linear and angular momentum ( see appendix a and b ) to write down the coupled nonlinear differential equations for the translational and rotational motions of the cylinder , as it sediments , slides and rolls down the incline : @xmath104-\cos\alpha\ , \nonumber\\
\ddot{x}_{\textrm{g}}&=&-\frac{2\epsilon\xi}{3}\ \frac{\dot{x}_{\textrm{g}}}{\sqrt{\delta}}-\frac{\kappa\epsilon\xi}{6}\left[\frac{19}{4}\frac{\dot{\delta}\dot{x}_{\textrm{g}}}{\delta^{7/2}}-\frac{\dot{\delta}\dot{\theta}}{\delta^{7/2}}+\frac12\frac{\ddot{\theta}-\ddot{x}_{\textrm{g}}}{\delta^{5/2}}\right]+\sqrt{\frac{\epsilon}{2}}\sin\alpha\ , \nonumber\\ \label{tdirect } \ddot{\theta}&=&-\frac{4\epsilon\xi}{3}\frac{\dot{\theta}}{\sqrt{\delta}}-\frac{\kappa\epsilon\xi}{3}\left[\frac{19}{4}\frac{\dot{\delta}\dot{\theta}}{\delta^{7/2}}-\frac{\dot{\delta}\dot{x}_{\textrm{g}}}{\delta^{7/2}}+\frac12\frac{\ddot{x}_{\textrm{g}}-\ddot{\theta}}{\delta^{5/2}}\right]\ .\end{aligned}\ ] ] we note that the lubrication pressure - induced torque vanishes since the corresponding forces act along the radii of the cylinder .
interestingly , this would not be the case for the opposite case of a soft cylinder which will deform asymmetrically near a rigid wall , thus breaking once a well admitted symmetry between the two dual systems in elasticity @xcite .
we see that particle inertia plays a central role in eq .
( [ tdirect ] ) , even though we have neglected fluid inertia . to justify this assumption ,
let us consider for instance a @xmath16-translation of the cylinder along the rigid wall , at typical speed @xmath105 and distance @xmath106 from the wall . in the navier - stokes equation ,
the local fluid inertia term reads @xmath107 , where @xmath108 is the typical time scale of the flow at speed @xmath105 , and @xmath109 the length of the contact zone along @xmath16 .
similarly , the convective inertia term reads @xmath110 . on the other hand ,
the viscous term in the navier - stokes equation reads @xmath111 .
the ratio of inertia over viscous terms thus reads @xmath112 , where the reynolds number is given by @xmath113 . as for particle inertia , following newton s law , we note that it scales as @xmath114 . according to our eq .
( [ 3drag ] ) , for the tangential motion along the rigid wall , the pressure - induced force ( per unit length ) scales as @xmath115 @xcite , in real variables .
the ratio of particle inertia and fluid viscosity thus reads @xmath116 , which is much larger than the ratio of fluid inertia and fluid viscosity due to the lubrication parameter @xmath117 even in the case when the densities are matched .
thus , we see that even if fluid inertia plays a role on short time scales , there is a range of parameters over which it is negligible while particle inertia is still important . this conclusion remains valid even in the presence of the additional compliance parameter @xmath84 describing the wall softness .
the elastohydrodynamic drag terms on the right - hand sides of eq .
( [ tdirect ] ) trigger an interesting zoology of solutions , which we now turn to .
the solutions are governed by four dimensionless control parameters corresponding to a ratio @xmath118 of viscous damping over gravitational driving , an incline angle @xmath119 , a scaled wall compliance @xmath120 , and a scaled lubrication gap @xmath117 .
in addition , there are three relevant initial conditions : @xmath121 , @xmath122 , and @xmath123 , since @xmath75 by virtue of our choice of the dimensionless variable @xmath124 , while all the initial tangential positions @xmath125 and initial angles @xmath126 are equivalent .
below , we give a brief flavour of some of the unexpected behaviours of the system with the aim of sketching the diversity of solutions , potentially valid for a variety of similar systems and experiments , rather than to build a complete phase - diagram for this 2d case . ) , for @xmath98 , @xmath127 , @xmath74 , @xmath75 , @xmath128 .
the dashed line corresponds to the case of a rigid wall , with @xmath129 , showing that @xmath103 decreases gradually , while the solid line corresponds to sedimentation towards a soft wall , with @xmath73 , where the cylinder abruptly crashes downwards after sedimenting at a rate faster than towards a rigid wall .
the figure on the right side shows that the horizontal position of the cylinder does not vary at all during sedimentation ( the simulation is extended to @xmath130 corresponding to the case of sedimentation towards a rigid wall).__,width=510 ] in fig .
[ fig2 ] , we show that when the cylinder is released along a steep incline , it slides along it uniformly even as it spontaneously oscillates , although these oscillations are damped .
indeed , the envelope decays over a dimensionless time that is consistent with our earlier scaling estimate : @xmath131 ( in dimensionless form ) for the parameters of fig .
similarly , the equilibrium height can be calculated by balancing gravity @xmath132 and the elastohydrodynamic lift @xmath133 in the first line of eq .
( [ tdirect ] ) , to yield : @xmath134 consistent with the dimensional scaling form given in eq .
( [ deltaeq ] ) ) . for the parameters in fig .
[ fig2 ] , one obtains @xmath135 , which is close to the observed average value of @xmath136 seen in fig .
[ fig2]a ; the slight difference is due to the weak influence of other terms in eq .
( [ tdirect ] ) . in fig .
[ fig3 ] , we show another peculiar effect associated with the case when the cylinder is started with an initial spin . as seen , it can lift off the soft wall via a magnus - like effect @xcite even as it slides along a horizontal wall .
this effect is due to the fluid shear induced by rotation that leads to an increased hydrodynamic pressure , which deforms the wall and thence leads to a normal force . ) .
the results follow from the numerical solution of eq .
( [ tdirect ] ) , for @xmath71 , @xmath127 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , @xmath137 ( red dots ) , and @xmath138 ( green lines ) . for these parameters , @xmath139 ( blue dash - dotted line ) , as obtained from eq .
( [ tresh]).__,width=510 ] next , we turn to examine the equations when the effective mass of the particle vanishes . indeed ,
since we kept both the cylinder inertia and the acceleration drag , as explained above , those two second - derivative terms may cancel each other .
this singularity leads to a vanishing effective mass and thus a diverging acceleration @xmath140 , and occurs at the three critical heights : @xmath141 in fig .
[ fig4 ] , we show the evolution of the height when a relatively heavy cylinder is released above a horizontal soft wall .
we see that the particle sediments is not exactly a cylinder - wall distance , since there is an additional @xmath142 term according to eq .
( [ gap ] ) .
however , this additional term can be computed from eq .
( [ tiltpres ] ) and the solutions of eq .
( [ tdirect ] ) , if needed . ] at an enhanced rate relative to the case when the soft wall is replaced by a rigid wall , corresponding to @xmath129 .
this is due to the fact that when the cylinder reaches the largest critical height @xmath143 of eq .
( [ crit ] ) , the vanishing effective inertial mass leads to an infinite acceleration .
this unphysical effect is regularized when one takes into account fluid inertia , leading to a smoothed out temporal profile of sedimentation . nonetheless ,
as explained above , the temporal cut - off associated with fluid inertia is assumed to correspond to time scales smaller than the ones associated with the motion of the cylinder , which means that , even if finite , @xmath140 may be large and the behaviour still very sharp .
finally , we note that when @xmath144 and @xmath145 , the singularity may instead occur at @xmath146 .
we now use our results to characterize the sedimentation - sliding transition for a cylinder falling down an incline .
equation ( [ splan ] ) suggests that the cylinder can stably slide at a dimensionless height @xmath147 . on the other hand , if for instance @xmath143 in eq .
( [ crit ] ) , @xmath148 fixes the relevant singular sedimentation height that may be encountered during the fall of the cylinder , as illustrated in fig . [ fig4 ] .
the balance of these two dimensionless heights yields the threshold tangential velocity @xmath149 above which sliding becomes possible : @xmath150 in fact , if @xmath151 , and thus @xmath152 , the singular sedimentation height @xmath148 is reached before the sliding height @xmath147 , and one typically gets sedimentation . if , in contrast , @xmath153 , and thus @xmath154 , the sliding height @xmath147 is reached before the singular sedimentation height @xmath148 , and one typically gets sliding .
this transition is illustrated in fig .
[ fig5 ] , for two given sets of dimensionless parameters and initial conditions .
for instance , with a meter - sized body , this reasonably corresponds to a @xmath155 threshold velocity .
we note that , although the presence of an elastic wall is crucial in the underlying mechanism , the elastohydrodynamic details do not appear in this purely gravitational expression . ) , for @xmath98 , @xmath99 , @xmath73 , @xmath74 , @xmath75 , @xmath100 , @xmath156 , and @xmath157.__,width=510 ] we conclude our tour of the zoology of solutions by noting that when a relatively heavy cylinder is released with spin and tangential velocity , it can reverse its direction of motion and return backwards along the soft wall , as shown in fig .
this effect can be understood by noting that the second equation in eq .
( [ tdirect ] ) characterizes the dynamics of sliding .
thus , when @xmath145 , a large enough positive spin velocity suffices to bring about a reversal in the tangential acceleration . once initiated and stabilized , the sliding motion eventually reaches a long - term steady - state , with a terminal velocity that reads : @xmath158 which is obtained by balancing viscous damping and gravity in the tangential component of eq .
( [ tdirect ] ) , and by replacing @xmath159 and @xmath147 with @xmath160 and @xmath161 in eq .
( [ splan ] ) , respectively .
this also leads to a prediction of the associated terminal sliding height : @xmath162 consistent with the scaling relations in eq .
( [ scaling1 ] ) .
the convergence to this long - term steady - state for the stable sliding case is illustrated by solving eq .
( [ tdirect ] ) and the results are depicted in fig .
[ fig7 ] , showing that the cylinder indeed reaches the terminal velocity and height obtained above .
naturally , these results are valid as long as @xmath103 remains sufficiently smaller than @xmath163 , so that the lubrication approximation holds .
this criterion corresponds to a terminal velocity @xmath160 being smaller than @xmath164 .
we conclude with a brief discussion of a generalisation of our results to the case when the wall is fluid permeable , a problem of some relevance to many biological and geological situations @xcite , and we follow and generalize the results of @xcite that we summarize below . ) .
the results follow from the numerical solution of eq .
( [ tdirect ] ) , for @xmath71 , @xmath127 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , and @xmath138 ( green lines ) . for those parameters , the terminal height and velocity of the sliding steady - state
are given by @xmath165 and @xmath166 ( orange dashed lines ) , as obtained from eqs .
( [ ssv ] ) and ( [ ssh ] ) respectively.__,width=510 ] we introduce the volume fraction @xmath167 of fluid in the porous wall , the bulk modulus @xmath168 of the solid porous matrix ( with @xmath8 now being the composite shear modulus of the poroelastic medium ) , and the isotropic darcy permeability @xmath169 , and we note that the pore size @xmath170 is small in comparison with the wall thickness @xmath7 .
we assume that there is no flow inside the poroelastic wall in comparison with the flow in the lubrication gap , which is valid as long as @xmath171 .
for example , if @xmath26 , this follows due to scale separation .
the fluid - permeable wall introduces a new time scale associated with flow - induced stress relaxation given by @xmath172 , which has to be compared with the lubrication time scale @xmath173 @xcite . if @xmath174 , the fluid in the wall is in equilibrium with the outside and a purely elastic theory suffices , so that eq . ( [ locel ] )
is modified to read : @xmath175 which simply corresponds to a small effective stiffening due to the presence of a volume fraction @xmath167 of fluid in the poroelastic wall .
in contrast , if @xmath176 , the pore fluid has no time to adapt and we find that the wall is effectively stiffer , with @xmath177 , so that : @xmath178 in both cases there is a purely local elastic response to the driving pressure field .
therefore , all our previous results directly apply to these limiting poroelastic cases as well , provided we use the transformations : @xmath179 if @xmath174 , and @xmath180 if @xmath176 .
using soft lubrication theory and scaling arguments , we have shown that when a cylinder moves freely close to an elastic or poroelastic wall , the flow - induced pressure field exerts a drag force that resists this motion , but it also deforms the wall , which may in turn increase the gap and reduce this drag , as well as create a supplementary lift .
this leads to a complex and rich zoology of inertial motions that link sedimentation , sliding , and spinning , despite the inertialess motion of the fluid .
indeed , it is the wall elasticity combined with the cylinder inertia that are at the origin of all these effects , even at low reynolds number .
the striking observed solutions include non - exhaustively : oscillations , magnus - like effect , spin - induced reversal , enhanced sedimentation , and long - term steady sliding .
while the fully three - dimensional motion of a sphere , or other solid , will have three additional degrees of freedom , we expect many of the qualitative scaling features that we have uncovered to persist .
in this first appendix , we detail the derivation of eq .
( [ tdirect ] ) at zeroth order in the dimensionless compliance @xmath84 of the substrate .
equation ( [ stokes ] ) is the stokes equation for the flow , and the no - slip boundary conditions read : @xmath181 and @xmath182 .
in addition , the profile of eq .
( [ gap ] ) becomes : @xmath183 ^ 2\ .\ ] ] the corresponding poiseuille velocity is thus given by : @xmath184+\frac{(\dot{x}_{\textrm{g}}+\dot{\theta})\ z}{\delta+\left(x - x_{\textrm{g}}\right)^2}\ .\ ] ] then , integrating once the volume conservation of eq .
( [ tfe ] ) , with respect to @xmath185 , leads to : @xmath186}{\left[\delta+\left(x - x_{\textrm{g}}\right)^2\right]^3}\ , \ ] ] where @xmath187 is an integration constant , that was identified thanks to the assumed vanishing lubrication pressure @xmath188 at @xmath189 . in this case
, a second spatial integration leads to : @xmath190 ^ 2}\ .\ ] ] the pressure is not an even function in @xmath185 due to the transverse motion , and therefore there is a tangential drag associated to it , in addition to the normal one .
we use eq .
( [ dragp ] ) to evaluate both projections . by parity ,
the total dimensionless pressure - induced drag force along @xmath191 is thus : @xmath192 similarly , the total dimensionless pressure - induced drag force along @xmath185 reads : @xmath193 which is smaller in magnitude by a factor @xmath194 than the orthogonal one along @xmath191 . it is important to highlight that we had to go to the next order in @xmath195 to obtain the pressure - induced drag force @xmath196 in the tangential direction , which is now of comparable magnitude as the tangential drag @xmath197 obtained from the dominant viscous stress component : @xmath198 .
therefore , one has to calculate the latter through eq .
( [ drags ] ) with @xmath129 : @xmath199 using eqs .
( [ tiltvel ] ) and ( [ tiltpres ] ) , it becomes : @xmath200 which precisely compensates the part of @xmath196 that depends on @xmath201 .
knowing the dominant drag in each direction , one can now study the motion of the cylinder in the presence of gravity and buoyancy .
the @xmath191-projection of the balance of linear momentum reads : @xmath202 thus , the sedimentation motion is decoupled from the others . in contrast , the sliding motion is coupled to the sedimentation motion through the @xmath185-projection of the balance of linear momentum , as given by : @xmath203 finally , the spinning motion can be obtained by the balance of angular momentum that reads : @xmath204 where the pressure - induced torque is zero since the pressure - induced force acts along a radius of the cylinder .
this can be non - dimensionalized as : @xmath205 which results in the trivial non - spinning solution , if @xmath102 , due to the absence of driving force .
this statement is modified for a soft wall , as studied below .
here , we detail the derivation of the central eq .
( [ tdirect ] ) at first order in the dimensionless compliance @xmath84 of the substrate . solving eq .
( [ stokes ] ) with the new boundary conditions : @xmath67 , and @xmath68 , and the gap profile of eq .
( [ gap ] ) , and conserving the volume of the fluid through eq .
( [ tfe ] ) , leads to eq .
( [ eqgen3 ] ) .
since @xmath206 depends on @xmath185 , a direct spatial integration of this equation would lead to an integro - differential equation .
we restrict ourselves to perturbation theory in @xmath120 , consistent with the assumption of linear elasticity : @xmath207 where both @xmath208 and @xmath209 are assumed to vanish at infinity .
equation ( [ eqgen3 ] ) at zeroth order in @xmath84 is equivalent to eq .
( [ eqspec2 ] ) , so that the zeroth order pressure follows from eq .
( [ tiltpres ] ) : @xmath210 ^ 2}\ , \ ] ] while the corresponding zeroth order drag forces from eqs .
( [ zeroperp ] ) , ( [ zeropara ] ) , and ( [ zerosig ] ) are : @xmath211 expressing eq .
( [ eqgen3 ] ) at first order in @xmath84 then yields : @xmath212_x=12p^{(0)}_t\ .\ ] ] proceeding as in appendix a , using three spatial integrations and the above - mentioned boundary conditions , one gets the normal and tangential pressure - induced drag forces as : @xmath213\ , \end{aligned}\ ] ] that are consistent with the steady - state results @xcite when @xmath214 and @xmath215 . in order to calculate the remaining first order viscous stress ,
one expresses the velocity gradient at the surface of the cylinder : @xmath216 where : @xmath217+\frac{p^{(0)}p^{(0)}_x}{2}-\frac{(\dot{\theta}+\dot{x}_{\textrm{g}})p^{(0)}}{\left[\delta+(x - x_{\textrm{g}})^2\right]^2}\ .\ ] ] therefore , using eq .
( [ drags ] ) , one gets : @xmath218\ .\ ] ] finally , the balance of linear and angular momentum leads to the general coupled system of three equations : @xmath219+\cos\alpha&=&0\ , \\
\label{xdirect } \ddot{x}_{\textrm{g}}+\frac{2\epsilon\xi}{3}\ \frac{\dot{x}_{\textrm{g}}}{\sqrt{\delta}}+\frac{\kappa\epsilon\xi}{6}\left[\frac{19}{4}\frac{\dot{\delta}\dot{x}_{\textrm{g}}}{\delta^{7/2}}-\frac{\dot{\delta}\dot{\theta}}{\delta^{7/2}}+\frac12\frac{\ddot{\theta}-\ddot{x}_{\textrm{g}}}{\delta^{5/2}}\right]-\sqrt{\frac{\epsilon}{2}}\sin\alpha&=&0\ , \\
\ddot{\theta}+\frac{4\epsilon\xi}{3}\frac{\dot{\theta}}{\sqrt{\delta}}+\frac{\kappa\epsilon\xi}{3}\left[\frac{19}{4}\frac{\dot{\delta}\dot{\theta}}{\delta^{7/2}}-\frac{\dot{\delta}\dot{x}_{\textrm{g}}}{\delta^{7/2}}+\frac12\frac{\ddot{x}_{\textrm{g}}-\ddot{\theta}}{\delta^{5/2}}\right]&=&0\ , \end{aligned}\ ] ] which corresponds to eq .
( [ tdirect ] ) . | we consider the motion of a fluid - immersed negatively buoyant particle in the vicinity of a thin compressible elastic wall , a situation that arises in a variety of technological and natural settings .
we use scaling arguments to establish different regimes of sliding , and complement these estimates using thin - film lubrication dynamics to determine an asymptotic theory for the sedimentation , sliding , and spinning motions of a cylinder .
the resulting theory takes the form of three coupled nonlinear singular - differential equations .
numerical integration of the resulting equations confirms our scaling relations and further yields a range of unexpected behaviours . despite the low - reynolds feature of the flow ,
we demonstrate that the particle can spontaneously oscillate when sliding , can generate lift via a magnus - like effect , can undergo a spin - induced reversal effect , and also shows an unusual sedimentation singularity .
our description also allows us to address a sedimentation - sliding transition that can lead to the particle coasting over very long distances , similar to certain geophysical phenomena .
finally , we show that a small modification of our theory allows to generalize the results to account for additional effects such as wall poroelasticity . |
ambitious projects have been launched to extend conventional astronomy beyond wavelengths of @xmath3 cm , or gev photon energy . besides gamma rays , protons ( nuclei ) ,
neutrinos and gravitational waves will be explored as astronomical messengers probing the extreme universe .
the challenges are considerable : * protons are relatively abundant , but their arrival directions have been scrambled by magnetic fields . * @xmath4-rays do point back to their sources , but are absorbed at tev - energy and above on cosmic background radiation . * neutrinos propagate unabsorbed and without deflection throughout the universe but are difficult to detect .
therefore , multi - messenger astronomy may not just be an advantage , it may be a necessity for solving some of the outstanding problems of astronomy at the highest energies such as the identification of the sources of the cosmic rays , the mechanism(s ) triggering gamma ray bursts and the particle nature of the dark matter .
we here discuss the detection of neutrinos associated with the observed fluxes of high energy cosmic rays and gamma rays .
we will show that the anticipated fluxes point at the necessity of commissioning kilometer - scale neutrino detectors .
though ambitious , the scientific case is compelling because neutrinos will reveal the location of the source(s ) and represent the ideal tool to study the black holes powering the cosmic accelerator(s ) .
soon after the discovery in the mid - fifties that neutrinos were real particles and not just mathematical constructs of theorists imagination , the idea emerged that they represent ideal cosmic messengers@xcite . because of their weak interactions , neutrinos reach us unimpeded from the edge of the universe and from the inner reaches of black holes .
the neutrino telescopes now under construction have the capability to detect neutrinos with energies from a threshold of @xmath5gev to , possibly , @xmath6eev , the highest energies observed .
their telescope range spans more than 10 orders of magnitude in wavelengths smaller than @xmath3 cm .
this is a reach equivalent to that of a hypothetical astronomical telescope sensitive to wavelengths from radio to x - rays .
above @xmath7tev the observations are free of muon and neutrino backgrounds produced in cosmic ray interactions with the earth s atmosphere .
each neutrino is a discovery.ev , tev@xmath8ev , pev@xmath9ev and eev@xmath10ev units of energy ] the real challenge of neutrino astronomy is that kilometer - scale neutrino detectors are required to do the science . the first hint of the scale of neutrino telescopes emerged in the nineteen seventies from theoretical studies of the flux of neutrinos produced in the interactions of cosmic rays with microwave photons , the so - called greissen - zatsepin - kuzmin or gzk neutrinos . since then the case for kilometer - size instruments has been strengthened@xcite and the possibility of commissioning such instruments demonstrated@xcite .
in fact , if the neutrino sky were within reach of smaller instruments , it would by now have been revealed by the first - generation amanda telescope .
it has been taking data since 2000 with a detector of 0.010.08km@xmath11 telescope area , depending on the sources@xcite . given the size of the detector required , all efforts have concentrated on transforming large volumes of natural water or ice into cherenkov detectors . they reveal the secondary muons and electromagnetic and hadronic showers initiated in neutrino interactions inside or near the detector . because of the long range of the muon , from kilometers in the tev range to tens of kilometers at the highest energies , neutrino interactions can be identified far outside the instrumented volume .
adding to the technological challenge is the requirement that the detector be shielded from the abundant flux of cosmic ray muons by deployment at a depth of typically several kilometers .
after the cancellation of a pioneering attempt@xcite to build a neutrino telescope off the coast of hawaii , successful operation of a smaller instrument in lake baikal@xcite bodes well for several efforts to commission neutrino telescopes in the mediterranean@xcite .
we will here mostly concentrate on the construction and first four years of operation of the amanda telescope@xcite which has transformed a large volume of natural deep antarctic ice into a cherenkov detector .
it represents a first - generation telescope as envisaged by the dumand collaboration over 20 years ago and a proof of concept for the kilometer - scale icecube detector , now under construction .
even though neutrino telescopes " are designed as discovery instruments covering a large dynamic range , be it for particle physics or astrophysics , their conceptual design is very much anchored to the observational fact that nature produces protons and photons with energies in excess of @xmath0v and @xmath1ev , respectively . in this paper
we will review how cosmic ray and tev gamma ray observations set the scale of cosmic neutrino fluxes .
cosmic accelerators produce particles with energies in excess of @xmath12tev ; we do not know where or how .
the flux of cosmic rays observed at earth is sketched in fig.1a , b@xcite .
the energy spectrum follows a broken power law .
the two power laws are separated by a feature dubbed the knee " ; see fig.1a .
circumstantial evidence exists that cosmic rays , up to perhaps eev energy , originate in galactic supernova remnants .
any association with our galaxy disappears in the vicinity of a second feature in the spectrum referred to as the ankle " . above the ankle ,
the gyroradius of a proton in the galactic magnetic field exceeds the size of the galaxy and it is generally assumed that we are witnessing the onset of an extragalactic component in the spectrum that extends to energies beyond 100eev .
experiments indicate that the highest energy cosmic rays are predominantly protons or , possibly , nuclei . above a threshold of 50 eev
these protons interact with cosmic microwave photons and lose energy to pions before reaching our detectors .
this is the gzk cutoff that limits the sources to our local supercluster .
models for the origin of the highest energy cosmic rays fall into two categories , top - down and bottom - up . in top - down models
it is assumed that the cosmic rays are the decay products of cosmological remnants or topological defects associated , for instance , with grand unified theories with unification energy @xmath13 .
these models predict neutrino fluxes most likely within reach of first - generation telescopes such as amanda , and certainly detectable by future kilometer - scale neutrino observatories@xcite . they have not been observed .
in bottom - up scenarios it is assumed that cosmic rays originate in cosmic accelerators . accelerating particles to tev energy and above requires massive bulk flows of relativistic charged particles .
these are likely to originate from the exceptional gravitational forces in the vicinity of black holes .
gravity powers large electric currents that create the opportunity for particle acceleration by shocks , a mechanism familiar from solar flares where particles are accelerated to @xmath14gev .
it is a fact that black holes accelerate electrons to high energy ; astronomers observe them indirectly by their synchrotron radiation .
we know that they can accelerate protons because we detect them as cosmic rays .
being charged , the protons are deflected by interstellar magnetic fields and therefore do not reveal their sources . hence the cosmic ray puzzle .
examples of candidate black holes include the dense cores of exploding stars , inflows onto supermassive black holes at the centers of active galaxies and annihilating black holes or neutron stars . before leaving the source ,
accelerated particles pass through intense radiation fields or dense clouds of gas surrounding the black hole .
this results in interactions producing pions decaying into secondary photons and neutrinos that accompany the primary cosmic ray beam as illustrated in fig.2 .
how many neutrinos are produced in association with the cosmic ray beam ?
the answer to this question provides one rationale for building kilometer - scale neutrino detectors @xcite . for orientation , consider a neutrino beam produced at an accelerator laboratory . here the target and the beam dump absorb all parent protons as well as the secondary electromagnetic and hadronic showers . only neutrinos exit the dump .
if nature constructed such a `` hidden source '' in the heavens , conventional astronomy would not reveal it .
cosmic ray sources must be at least partially transparent to protons .
sources transparent only to neutrinos may exist , but they can not be cosmic - ray sources .
a generic `` transparent '' source can be imagined as follows : protons are accelerated in a region of high magnetic fields where they interact with photons and generate neutral and charged pions .
the most important process is @xmath15 and @xmath16 . while the secondary protons may remain trapped in the acceleration region , roughly equal numbers of neutrons and decay products of neutral and charged pions escape .
the energy escaping the source is therefore distributed among cosmic rays , gamma rays and neutrinos produced by the decay of neutrons , neutral pions and charged pions , respectively .
the neutrino flux from a generic transparent cosmic ray source is often referred to as the waxman - bahcall flux @xcite .
it is easy to calculate and the derivation is revealing .
figure 1b shows a fit to the observed spectrum above the ankle " that can be used to derive the total energy in extragalactic cosmic rays . the flux above the ankle is often summarized as one @xmath17ev particle per kilometer square per year per steradian " .
this can be translated into an energy flux @xmath18 & = & 3\times 10^{-8}\rm\ , gev\ cm^{-2 } \ ,
s^{-1 } \ , sr^{-1 } \,.\end{aligned}\ ] ] from this
we can derive the energy density @xmath19 in cosmic rays using the relation that flux@xmath20velocity@xmath21density , or @xmath22 we obtain @xmath23 taking the extreme energies of the accelerator(s ) to be + @xmath24 .
the energy content derived professionally " by integrating the spectrum in fig .
2b assuming an @xmath25 energy spectrum , typical of shock acceleration , with a gzk cutoff is @xmath26 .
this is within a factor of our back - of - the - envelope estimate ( 1tev = 1.6erg ) .
the power required for a population of sources to generate this energy density over the hubble time of @xmath27years is @xmath28 per ( mpc)@xmath29 or , as often quoted in the literature , @xmath30 per ( mpc)@xmath29 per year .
this works out to@xcite * @xmath31 per galaxy , * @xmath32 per cluster of galaxies , * @xmath33 per active galaxy , or * @xmath34erg per cosmological gamma ray burst .
the coincidence between these numbers and the observed output in electromagnetic energy of these sources explains why they have emerged as the leading candidates for the cosmic ray accelerators .
the coincidence is consistent with the relationship between cosmic rays and photons built into the transparent " source . in the photoproduction
processes roughly equal energy goes into the secondary neutrons , neutral and charged pions whose energy ends up in cosmic rays , gamma rays and neutrinos , respectively .
we therefore conclude that the same energy density of @xmath35 , observed in cosmic rays and electromagnetic energy , ends up in neutrinos with a spectrum @xmath36 that continues up to a maximum energy @xmath37 .
the neutrino flux follows from the relation @xmath38 . for @xmath39 and @xmath40gev ,
the generic source of the highest energy cosmic rays produces a flux of @xmath41 .
there are several ways to sharpen this qualitative prediction : * the derivation fails to take into account that there are more uhe cosmic rays in the universe than observed at earth because of the gzk - effect and it also neglects the evolution of the sources with redshift .
this increases the neutrino flux , which we normalized to the observed spectrum only , by a factor @xmath42 , the ratio of the hubble radius to the average attenuation length of the cosmic rays propagating in the cosmic microwave background . * for proton-@xmath4 interactions muon neutrinos ( and antineutrinos ) receive only 1/2 of the energy of the charged pion in the decay chain @xmath43 assuming that the energy is equally shared between the 4 leptons .
furthermore half the muon neutrinos oscillate into tau neutrinos over cosmic distances .
in further calculations we will focus on the muon flux here . in summary , @xmath44 in practice ,
the corrections approximately cancel .
the precise value of the energy where the transition from galactic to extragalactic sources occurs represents another source of uncertainty that has been extensively debated @xcite .
a transition at a lower energy significantly increases the energy in the extragalactic component and results in an enhancement of the associated neutrino flux .
waxman and bahcall referred to their flux as a bound in part because in reality more energy is transferred to the neutron than to the charged pion in the source , in the case of the photoproduction reaction @xmath45 four times more .
therefore @xmath46 in the end we estimate that the muon - neutrino flux associated with the sources of the highest energy cosmic rays is loosely confined to the range @xmath47 depending on the cosmological evolution of the cosmic ray sources . model calculations assuming that active galaxies or gamma - ray bursts are the actual sources of cosmic rays yield event rates similar to the generic energetics estimate presented .
the anticipated neutrino flux thus obtained has to be compared with the limit of + @xmath48 reached after the first 4 years of operation of the completed amanda detector in 20002003 @xcite . on the other hand , after three years of operation icecube will reach a diffuse flux limit of @xmath49 the exact value of the icecube sensitivity depends on the magnitude of the dominant high energy neutrino background from the prompt decay of atmospheric charmed particles@xcite .
the level of this background is difficult to anticipate theoretically and little accelerator data is available in the energy and feynman - x range of interest@xcite .
the observed event rate is obtained by folding the cosmic flux predicted with the probability that the neutrino is actually detected in a high energy neutrino telescope ; only one in a million neutrinos of tev energy interact and produce a muon that reaches the detector .
this probability is given by the ratio of the muon and neutrino interaction lengths in the detector medium , @xmath50@xcite and therefore depends on energy . for the flux range estimated above
we anticipate 100500 detected muon neutrinos per km@xmath11 per year . here
the lower value represents the more realistic estimate .
it will be further reduced if we assume a steeper spectrum . on the other hand , given that icecube s effective area for muon neutrinos exceeds 1km@xmath11 and that equal fluxes of electron and tau neutrinos are expected , a neutrino signal at the waxman - bahcall " level could result in the observation of several hundred high - energy neutrinos of extraterrestrial origin per year in icecube @xcite .
gamma ray bursts ( grb ) , outshining the entire universe for the duration of the burst , are perhaps the best motivated sources of high - energy neutrinos@xcite .
the collapse of massive stars to a black hole has emerged as the likely origin of the long " grb with durations of tens of seconds . in the collapse
a fireball is produced which expands with a highly relativistic velocity powered by radiation pressure .
the fireball eventually runs into the stellar material that is still accreting onto the black hole .
if it successfully punctures through this stellar envelope the fireball emerges to produce a grb .
while the energy transferred to highly relativistic electrons is thus observed in the form of radiation , it is a matter of speculation how much energy is transferred to protons .
the assumption that grb are the sources of the highest energy cosmic rays does determine the energy of the fireball baryons . accommodating the observed cosmic ray spectrum of extragalactic cosmic rays requires roughly equal efficiency for conversion of fireball energy into the kinetic energy of protons and electrons . in this scenario
the production of neutrinos of 1001000tev energy in the grb fireball is a robust prediction because neutrinos are inevitably produced in interactions of accelerated protons with fireball photons .
estimates of the flux@xcite point again at the necessity of a kilometer - cubed neutrino detector , in agreement with the generic energetics estimates previously presented .
studies of active galaxies as sources of cosmic rays lead to similar conclusions@xcite .
the case for kilometer - scale detectors also emerges from consideration of `` guaranteed '' cosmic fluxes .
neutrino fluxes are guaranteed when both the accelerator and the pion producing target material can be identified .
the extragalactic cosmic rays produce @xmath51 1 event per km@xmath11year in interactions with cosmic microwave photons@xcite .
galactic cosmic rays interact with hydrogen in the disk to generate an observable neutrino flux in a kilometer - scale detector@xcite .
evidence has been accumulating that young supernova remnants are the sources of the galactic cosmic rays ; conclusive evidence is still missing .
neutrino observations can be the answer as we will review in the next section .
in the previous section we made an estimate of the neutrino flux from generic accelerators producing the highest energy cosmic rays . we can perform a similar analysis for the galactic cosmic rays by calculating the energy density corresponding to the flux shown in fig.1a .
the answer is that @xmath52erg@xmath53 .
this is also the value of the corresponding energy density @xmath54 of the microgauss magnetic field in the galaxy .
the power needed to maintain this energy density is @xmath55erg / cm@xmath29s given that the average containment time of the cosmic rays in our galaxy is @xmath57years .
for a nominal volume of the galactic disk of @xmath58cm@xmath29 this requires an accelerator delivering @xmath59erg / s .
this happens to be 10% of the power produced by supernovae releasing @xmath60erg every 30 years .
the coincidence is the basis for the idea that shocks produced by supernovae expanding into the interstellar medium are the origin of the galactic cosmic rays .
can we observe neutrinos pointing back at the accelerators of the galactic cosmic rays ?
the conversion of the @xmath61erg of energy into particle acceleration is believed to occur by diffusive shock acceleration in the young ( 100010,000 years ) remnant expanding into the interstellar medium .
if high energy cosmic rays are indeed associated with the remnant , they will interact with hydrogen atoms in the interstellar medium to produce pions that decay into roughly equal numbers of photons and neutrinos .
these may provide us with indirect evidence for cosmic ray acceleration .
the observation of these pionic gamma rays has been one of the motivations for neutrino as well ground - based tev - energy astronomy .
whereas the details are complex and predictions can be treacherous , a simple estimate of the gamma ray flux associated with a supernova remnant can be made following aharonian , drury and volk@xcite .
within the precision of the astrophysics it is safe to assume that an identical flux of neutrinos is produced no need for sophistication here .
the emissivity in pionic gamma rays produced by a density of protons @xmath62 interacting with a density of hydrogen atoms n is @xmath63 or @xmath64 the emissivity of photons is simply proportional to the density of cosmic rays @xmath65 ( @xmath66 for energy in excess of 1 tev ) and the target density @xmath67 of hydrogen atoms .
the proportionality factor is determined by particle physics : @xmath68 is the average energy of the secondary pions relative to the cosmic ray protons and @xmath69 is the proton interaction length ( @xmath70 mb ) in a density n of hydrogen atoms .
( we here assumed a generic @xmath25 spectrum of the protons , for different spectral indices the quantity @xmath71 is generalized to the spectrum - weighted moments for pion production by nucleons@xcite . )
the total luminosity in gamma rays is given by @xmath72 the density of protons from a supernova converting a total kinetic energy w of @xmath73erg to proton acceleration is approximately given by @xmath74 , where we will assume that the density in the remnant is not very different from the ambient energy density @xmath75erg@xmath53 of galactic cosmic rays .
this approximation is valid for young remnants in their sedov phase .
we thus predict a rate of tev photons from a supernova at a distance d of 1kpc of @xmath76 each tev gamma ray is accompanied by a neutrino from a charged pion and we therefore anticipate an event rate of 3 detected neutrinos per decade of energy per km@xmath11 year , a result readily obtained from the relation @xmath77 where the last factor represents , as before , the probability that the neutrino is detected .
it is approximately @xmath78 for the tev energy considered here . from several such sources distributed over the galactic plane icecube
may detect a flux of neutrinos similar , possibly smaller , than the one associated with extragalactic sources .
this estimate may be somewhat optimistic because we assumed that the sources extend to 100tev with an @xmath25 spectrum . if the spectrum cuts off around 10 tev detection
becomes more challenging because the flux reaches the level of the cosmic ray background . on the other hand , if the knee " at 1000tev represents the end of the galactic cosmic ray spectrum , then some of the sources must produce 100tev secondaries .
this prediction is credible because the number of tev photons predicted coincides with observations of the supernova remnant rx j1713.7 - 3946 by the h.e.s.s .
array of atmospheric cherenkov telescopes@xcite .
may thus have identified the first site where protons are accelerated to energies typical of the main component of the galactic cosmic rays .
although the resolved image of the source ( the first ever at tev energies ! ) reveals tev gamma ray emission from the whole supernova remnant , it shows a clear increase of the flux in the directions of known molecular clouds .
this is suggestive of protons , shock accelerated in the supernova remnant , interacting with the dense clouds to produce neutral pions that are the source of the observed increase of the tev photon signal .
the image shows filaments of high magnetic fields consistent with the requirements for acceleration to the energies observed .
furthermore , the high statistics data for the flux are power - law behaved over a large range of energies without any indication of a cutoff characteristic of synchrotron or inverse - compton sources .
follow - up observations of the source in radio - waves and x - rays have failed to identify the population of electrons required to generate tev photons by purely electromagnetic processes ; for a detailed discussion see @xcite . on the theoretical side ,
the large b - fields suppress the ratio of photons produced by the inverse compton relative to the synchrotron .
fitting the data by purely electromagnetic processes is therefore challenging but , apparently , not impossible@xcite .
a similar extended source of tev gamma rays tracing the density of molecular clouds has been identified near the galactic center .
protons apparently accelerated by the remnant hess j1745 - 290 diffuse through nearby molecular clouds to produce a signal of tev gamma rays that trace their density@xcite .
detecting this source in neutrinos will be challenging because it is relatively weak ( its tev luminosity is only or order 0.1 crab ) , because of its larger distance compared to rx j1713.7 - 3946 and because it is not a point but extended source@xcite . on the other hand , the sources discovered by milagro in the cygnus region are more luminous , their spectrum extends to higher energies and they are relatively nearby@xcite .
we are looking forward to a detailed measurement of the milagro spectrum which is likely to translate into a detectable neutrino flux in icecube .
the hotspot in the milagro map of the galactic plane represents a flux of 1 crab above 12.5tev@xcite . by itself
it yields @xmath79 neutrino events above a detector threshold of @xmath80tev .
we here assumed a spectrum of @xmath81 , but the result is relatively insensitive to this assumption .
the diffuse flux surrounding the milagro source represents an additional flux of 3 crab spread over 0.02 steradian .
icecube data should have no problem exposing the smoking gun for the sources of galactic cosmic rays .
so far h.e.s.s .
has not claimed the discovery of pionic gamma rays and finding neutrinos as a smoking gun for cosmic ray acceleration in supernova remnants remains of interest .
if the tev flux of rx j1713.7 - 3946 is of neutral pion origin , then the accompanying charged pions will produce a guaranteed neutrino flux of roughly 10 muon - type neutrinos per kilometer - squared per year@xcite and produce incontrovertible evidence for cosmic ray acceleration .
their calculation yields a result close to our previous estimate for a generic remnant . from a variety of such sources
we can therefore expect event rates of cosmic neutrinos of galactic origin similar to those estimated for extragalactic neutrinos in the previous section .
supernovae associated with molecular clouds are a common feature of associations of ob stars that exist throughout the galactic plane , e.g. in the cygnus region within view of icecube .
it is important to realize that there is a robust relation between the neutrino and gamma flux emitted by cosmic ray accelerators@xcite .
it can also be exploited to estimate the neutrino flux from extragalactic sources .
the @xmath82 neutrino flux ( @xmath83 ) produced by the decay of charged pions in the source can be derived from the observed gamma ray flux by energy conservation : @xmath84 where @xmath85 ( @xmath86 ) is the minimum ( maximum ) energy of the photons that have a hadronic origin .
@xmath87 and @xmath88 are the corresponding minimum and maximum energy of the neutrinos .
the factor @xmath89 depends on whether the @xmath90 s are of @xmath91 or @xmath92 origin .
its value can be obtained from routine particle physics . in @xmath91 interactions
1/3 of the proton energy goes into each pion flavor . in the pion - to - muon - to - electron decay chain 2 muon - neutrinos
are produced with energy @xmath93 for every photon with energy @xmath94 .
therefore the energy in neutrinos matches the energy in photons and @xmath95 .
the flux has to be reduced by a factor 2 because of oscillations . for @xmath92 interactions @xmath96 .
the estimate should be considered a lower limit because the observed photon flux to which the calculation is normalized may have been attenuated by absorption in the source or in the interstellar medium . in summary , the energetics of galactic as well as extragalactic cosmic rays
points at the necessity to build kilometer - scale detectors to observe the associated neutrino fluxes that will reveal the sources .
the case for doing neutrino astronomy with kilometer - scale instruments can also be made in other ways@xcite and , as is usually the case , the estimates of the neutrino fluxes pointing at the necessity of such detectors are likely to be optimistic .
i thank my icecube collaborators as well as luis anchordoqui , julia becker , concha gonzalez - garcia , tom gaisser , aongus omurchadha and grant teply for discussions .
this research was supported in part by the national science foundation under grant no .
opp-0236449 , in part by the u.s .
department of energy under grant no .
de - fg02 - 95er40896 , and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation .
99 k. greisen , ann . rev
. nucl . part .
sci . * 10 * , 63 ( 1960 ) , m.a.markov in proceedings of the 1960 international conference on high energy physics , e.c.g .
sudarshan , j.h.tinlot and a.c.melissinos editors , 578 ( 1960 ) .
t. k. gaisser , f. halzen , and t. stanev , phys .
rept . * 258 * , 173 ( 1995 ) [ erratum * 271 * , 355 ( 1995 ) ] , hep - ph/9410384 ; j.g . learned and k. mannheim , ann .
science * 50 * , 679 ( 2000 ) ; f. halzen and d. hooper , rept .
* 65 * , 1025 ( 2002 ) , arxiv : astro - ph/0204527
. a achtenberg _ et al .
_ , first year performance of the icecube neutrino telescope , arxiv : astro - ph/0604450 ; j. ahrens et al .
( icecube collaboration ) , astropart . phys . * 20 * , 507 ( 2004 ) , arxiv : astro - ph/0305196 and http://icecube.wisc.edu .
a. achterberg et al . , proceedings of the 29th international cosmic ray conference , pune , india , 2005 , arxiv : astro - ph/0509330 ; g. hill , invited talk at the same conference .
j. babson _
_ , dumand collaboration , phys .
d * 42 * , 3613 ( 1990 ) . v. a. balkanov et al .
( baikal collaboration ) , nucl .
. suppl .
* 118 * , 363 ( 2003 ) .
e. migneco _ et al .
_ , nucl .
* 136 * , 61 ( 2004 ) .
a. karle for the amanda collaboration , _ observation of atmospheric neutrino events with amanda _ , proceedings of the 26th international cosmic ray conference , salt lake city , utah ( 1999 ) ; e. andres et al .
( amanda collaboration ) , nature * 410 * , 441 ( 2001 ) ; phys.rev.d * 66 * , 012005 ( 2002 ) , arxiv : astro - ph/0205109 .
t. k. gaisser , proceedings of the 31st international conference on high energy physics , amsterdam , the netherlands , july 2002 .
d. v. semikoz and g. sigl , jcap * 0404 * , 003 ( 2004 ) , arxiv : hep - ph/0309328 . j. n. bahcall and e. waxman , phys .
d * 64 * , 023002 ( 2001 ) .
f. w. stecker and m. h. salamon , astrophys .
j. * 512 * , 521 ( 1992 ) , arxiv : astro - ph/9808110 ; a. atoyan and c. d. dermer , phys .
* 87 * , 221102 ( 2001 ) , arxiv : astro - ph/0108053 and references therein .
f. w. stecker , phys .
d. * 72 * , 107301 ( 2005 ) , arxiv : astro - ph/0510537 for a recent update . | although kilometer - scale neutrino detectors such as icecube are discovery instruments , their conceptual design is very much anchored to the observational fact that nature produces protons and photons with energies in excess of @xmath0ev and @xmath1ev , respectively .
the puzzle of where and how nature accelerates the highest energy cosmic particles is unresolved almost a century after their discovery . from energetics considerations we anticipate order @xmath2 neutrino events per kilometer squared per year pointing back at the source(s ) of both galactic and extragalactic cosmic rays . in this context , we discuss the results of the amanda and icecube neutrino telescopes which will deliver a kilometer - square - year of data over the next 3 years . |
the glass transition and slow glassy dynamics are phenomena most widely studied in the context of supercooled liquids and polymer melts angel96,ang88 .
important progress has been made in understanding the complexity of these phenomena by combining dynamical approaches with the concept of an underlying energy landscape , as is discussed in detail in other chapters of this volume .
for example , on the level of the mean field theory of glasses , it has been demonstrated that the dynamical , ideal mode coupling theory @xcite and energy landscape based replica mean field theories describe the same underlying physics , yet from rather different perspectives @xcite .
novel replica approaches have been developed to characterize the emergence of a metastable amorphous solid mon95,franz95,mepa98.1,mepa98.2 .
in addition , droplet arguments have been proposed to include important physics beyond the pure mean field description and led to the formulation of the random first order transition theory of glasses @xcite .
the latter is particularly important if one wants to make specific predictions for experiments that require one to go beyond the mean field limit @xcite .
this random first order transition theory of glasses with an underlying entropy crisis @xcite offers a general and quantitative description of structural glasses .
in addition , it defines a universality class for complex many body systems that is of importance for a much larger class of materials . .
for a specific case of a structural glass @xmath0 plays a role of density of the local atomic configurations .
glassy behavior is characterized by the emergence of large number of metastable states separated by high energy barriers . ]
glassy dynamics usually occurs when a system is supercooled below a first order transition to an ordered state . in supercooled liquids
this ordered state is the crystalline solid .
if one applies the same ideas to systems with charge or spin collective degrees of freedom in a correlated material , the ordered state corresponds to an electron crystal or an ordered magnetic state .
the equilibrium state is therefore ordered .
supercooling the liquid state is possible once the nucleation barriers are high .
thus , in what follows we ignore this ordered state and consider time scales that are short compared to the nucleation time @xmath1 .
one can now ask the question whether and for how long one can still consider the supercooled liquid in local equilibrium . here , by local equilibrium we mean that the system is ergodic and all configurations are being explored with a probability given by the boltzmann distribution ( only excluding the sharp peak that corresponds to the ordered state ) .
one expects that this supercooled state will fall out of equilibrium over very long time scales , i.e. behave nonergodically , once the barriers between distinct configurations of comparable energy become very large , fig .
if this happens , the system will explore only a subset of the available phase space and will eventually freeze .
suppose there are @xmath2 of such metastable configurations , that are separated by high barriers , fig .
[ fig2 ] . a calorimetric measurement
will then yield an entropy that is reduced by @xmath3compared to the ergodic situation , where @xmath4 is referred to as the configurational entropy .
if @xmath2 is exponentially large in the system size , @xmath4 becomes extensive and the dynamically frozen state strictly speaking does not obey the third law of thermodynamics .
for liquids it appears that @xmath5 on infinite time scales would extrapolate to zero at a temperature @xmath6 .
extrapolating further to @xmath7 would give @xmath5 less then zero .
this is the entropy crisis of kauzmann [ 16 ] .
such a crisis is avoided by an ideal glass transition in the models highlighted here .
a crucial question to explore is under what circumstances the configurational entropy is a well defined and meaningful concept .
metastable states become sharp conceptually only within a mean field theory description , where barriers can be infinitely high and a system might get trapped in a local minimum of phase space forever . within mean field theory , the number of states that are separated by high barriers proliferates at the mode coupling temperature @xmath8 . the sudden onset of @xmath9 at @xmath10
should not be misunderstood to mean that new configurations emerge at that temperature .
these configurations were present already in the liquid state above @xmath8 .
what happens at @xmath8 is that the barriers grow and that the system s dynamics becomes _ localized in configuration space_. it is this localization in configuration space that is the key element of the random first order theory ( not the ultimate glass transition ) . within mean field theory
the system freezes right at @xmath8 , while activated events , that are a key element of the theory , allow for a slowed down dynamics even below @xmath8 , reflecting the fact that those barriers are large but finite .
one might criticize this approach since going beyond mean field theory does not allow for a sharp thermodynamic definition of @xmath4 .
however , there are well known examples that demonstrate the opposite : understanding overheated or supercooled phases close to a first order transition also starts from the mean field concept of a local minimum in the energy landscape . once supplemented with an appropriate droplet theory of the metastable state it yields a description of the nucleation dynamics at first order transitions@xcite .
the logic used in the random first order theory of glasses is very similar .
one first develops a mean field theory and then supplements it by an appropriate droplet calculation that clearly goes beyond strict mean field theory .
there are numerous indications that the underlying principles that govern glassy behavior in supercooled liquids apply to other many body systems as well .
generally , glassy systems are characterized by slow relaxations and a broad spectrum of excitations , a behavior found in a number of strongly correlated hard condensed matter physics systems .
a particularly interesting situation occurs when interactions , that change very differently with distance , compete with one another . in such
_ uniformly frustrated systems _ an intermediate length scale can emerge that leads to new spatial structures and inhomogeneities . a classical example is a ferromagnet where finite range exchange interactions compete with the long range dipole - dipole interactions and leads to the formation of magnetic domains .
similar behavior is in fact abundant in a number of correlated electron systems .
examples are stripe formation in doped mott insulators stripe , stripe2,mang01,millis96,dagotto , nmr01,nmr02,nmr03,nmr3b , nmr04,msr01,msr02 , defect formation in two - dimensional electron systems popovich07,spivak2006,2dspivak2010 , bubbles of electronic states of high landau levels in quantum hall systems @xcite , magnetic domains in magnetic multilayer compounds @xcite , and mesoscopic structures formed in self - assembly systems @xcite .
these systems typically exhibit a multi - time - scale dynamics similar to the relaxation found in glasses .
this observation suggests that glassy behavior and large relaxation times are caused by the competition of interactions with different characteristic length scales popovic2002,popovic2,vlad2002,denis2002,ohta86,kivelson , nussinov , schmalian , sww00,wsw03,wsw04 , and not primarily due to the presence of strong disorder in the system . as we will see
, this does not mean that disorder is irrelevant for uniformly frustrated systems .
in fact , the opposite is true .
we find that even the smallest amounts of disorder and imperfections drive such system into a non - ergodic regime that otherwise emerges only in the presence of very strong disorder . in the language of a renormalization group flow
this means that the strength of disorder , @xmath11 , flows to larger and larger values and reaches a strong - disorder fixed point @xmath12 , even if the physical , bare disorder strength is very small .
glassiness arises spontaneously even for infinitesimal extrinsic disorder , leading to _ self - generated randomness _ ,
[ fig3 ] . in
uniformly frustrated systems self generated randomness and the emergence of an entropy crisis were first discussed in ref .
for a beautiful example of self generated randomness in the context of coupled josephson junction arrays , see ref .
@xcite . while frustration through competing interactions is easiest to analyze , the notion that uniform geometric frustration is at the heart of the structural glass transition was already discussed and analyzed in refs.sadoc,nelson,sethna with the underlying view that one can capture frustration through an underlying ideal structure in curved space as a reference state .
recently this general idea was analyzed in the specific context of glass formation in a hyperbolic space@xcite .
modeling frustrated interactions by a competing interactions on different length scales was suggested in ref.@xcite where the concept of avoided criticality for uniformly frustrated systems was introduced , see also refs . @xcite for further details .
evidence for fragile glass - forming behavior in the relaxation of coulomb frustrated three - dimensional systems by means of a monte carlo simulation was given in ref.@xcite .
furthermore , in ref.@xcite the close relation between the replica approach presented here and the results obtained within a mode coupling analysis of a uniformly frustrated system , was demonstrated explicitly .
for a recent review on the subject , see ref .
@xcite in this chapter we discuss in some detail model systems that display self generated randomness .
we do so by applying the replica formalism originally developed to describe disordered spin glasses @xcite and structural glasses @xcite .
we demonstrate that there are many body systems that undergo a mean field glass transition for arbitrarily weak disorder @xcite .
we then develop a landau theory of the glassy state that reproduces the key physical behavior of these uniformly frustrated systems @xcite .
the appeal of the landau theory is that it easily permits for a generalization beyond the mean field limit , where we include instanton events to describe dynamical heterogeneity in glassy systems .
we summarize the main featured of a simple model that exhibits glassy behavior due to infinitesimal amount of disorder schmalian , sww00,wsw03,wsw04 .
the model exhibits competition of interactions on different length scales and there are no explicitly quenched degrees of freedom .
consider @xmath13 : the spatially varying amplitude that characterizes the collective degrees of freedom of a many body system .
depending on the problem under consideration @xmath14 corresponds to the magnetization @xmath15 ( in case of magnetic domain formations ) or a density fluctuation relative to the mean density , @xmath16 . here , @xmath17 stands for the electron density in case of a doped mott insulator @xcite or for the density of amphiphilic molecules in case of a microemulsion or a copolymer leibler , fredrickson , deem94,wwsw02,zhang06,wu09 .
we consider the effective hamiltonian@xmath18where @xmath19 contains all terms of second order in @xmath13 .
specifically , we consider @xmath20 ^{2}\right ) \notag \\ & & + \frac{1}{2}\int d^{d}x\int d^{d}x^{\prime } \varphi ( \mathbf{x)}v\left ( \left\vert \mathbf{x - x}^{\prime } \right\vert \right ) \varphi ( \mathbf{x}^{\prime } ) .
\label{h0}\end{aligned}\]]for the non - linear interaction term we assume the local form @xmath21where we use @xmath22 in eq .
( [ h0 ] ) , @xmath23 favors local order of @xmath24 .
thus , in the absence of the long distance coupling @xmath25 the system is expected to be homogeneously ordered .
the gradient term favors homogeneous configurations of @xmath26 .
it is written by taking the the continuum limit of an underlying short range interaction . in the case where @xmath27 refers to the deviation from the mean density ,
the average @xmath28 vanishes by construction .
still the expected low temperature minimum of the free energy corresponds to configurations where regions of macroscopic size have constant field values ( @xmath29 ) , separated by domain walls of minimal area , due to the presence of the gradient term .
this is in fact a toy model for macroscopic phase separation .
the corresponding ordering temperature can be estimated within mean field theory .
for @xmath30 it follows , for example , @xmath31 , with momentum cutoff @xmath32 of the order of an inverse lattice constant .
the situation changes significantly once we include an additional long range interaction @xmath33here @xmath34 is the new typical length scale that results from the competition of the gradient term and the long range interaction with @xmath35 .
the long range interaction vanishes in the limit @xmath36 .
the exponent @xmath37 determines the rate of decay of the interaction .
the effect of this long range interaction can easily be seen in momentum space .
the fourier transform of @xmath38 is given as @xmath39with dimensionless coefficient @xmath40 .
for @xmath41 , homogeneous field configurations with @xmath42 are energetically very costly and spatial pattern with finite wave number @xmath43 emerge as consequence of the competition between the short range interaction and long range forces . for @xmath44 , @xmath45 and homogeneous configurations
are still suppressed , albeit only logarithmically .
this is relevant for the dipole - dipole interactions in @xmath30 .
one can also consider situations with a screened interaction , such as @xmath46as shown in ref .
@xcite , glass formation is virtually unchanged compared to the unscreened interaction once @xmath47 , while it is suppressed in the opposite limit .
the theory of refs .
@xcite was also applied to study the glassy phase in a model for charged colloids @xcite . at the hartree level ,
the correlation function that results from these two competing interactions is @xmath48with @xmath49 .
the intermediate length scale that results from the competition between short and long range interactions leads to a peak in the correlation function at @xmath50where @xmath51 .
we can approximate @xmath52 close to @xmath53 as : @xmath54where @xmath55 .
the liquid state correlation length is given by @xmath56 .
its temperature dependence must be determined from a conventional equilibrium calculation
. the form eq .
( [ braz ] ) holds for a large class of models with competing interactions . for
@xmath57 brazovskii showed that such a model undergoes a fluctuation induced first order transition to a state with lamellar , or stripe order @xcite . in general
, this model has been discussed in the context of complex crystallization @xcite .
for example , alexander and mctague @xcite argued that crystallization of body centered cubic crystals is preferred if the first - order character is not too pronounced .
as shown in refs .
groh99,klein01 the preference for bcc order is rather for metastable states that form near the spinodal .
below we will ignore ordered crystalline states and assume that their nucleation kinetics is sufficiently slow to supercool the disordered state .
finally , we mention that our model , eqs .
( [ h]-[hint ] ) , can also be used to describe interacting liquids consisting of @xmath58-components wsw04 . in this case
@xmath59 is a vector whose components refer to the density deviations of the different components of the liquid , with mean density @xmath60 .
now , the gaussian part @xmath61 \varphi _ { l^{\prime } } \left ( \mathbf{x}^{\prime } \right ) \end{split}\]]is determined by the direct correlation function @xmath62 of the fluid @xcite . within the density functional approach pioneered by ramakrishnan and youssoff @xcite the nonlinear part of the hamiltonian is determined by the ideal gas free energy : @xmath63 -\frac{\varphi _ { l}{}^{2}}{2\rho _ { 0,l}}\right]\]]in case of the density functional theory , eq .
( [ exp ] ) corresponds to the resummed virial expansion of the liquid . in summary ,
models of the form eqs .
( [ h]-[hint ] ) capture the physics of a large class of systems where competing interactions on different length scales lead to complex spatial pattern and potentially to slow glassy dynamics . in the next section we discuss in some detail the methods that will be used to describe and
analyze glassy systems .
those methods will then be applied to the model of eqs .
( [ h]-[hint ] ) further below .
we first outline the main idea of the generalized replica mean field approach for systems with an entropy crisis @xcite .
we start from a classical field theory with hamiltonian , @xmath64 $ ] , and field variable , @xmath26 , that yields the partition function @xmath65\right ) .
\label{psum_class}\]]the thermodynamic free energy is then given as @xmath66 .
suppose we know the thermodynamic density of states @xmath67\right ) , \]]where @xmath68 is the system volume .
it follows with @xmath69 that @xmath70 .\]]for large @xmath68 the integral is evaluated at the saddle point , leading to the free energy density @xmath71 .\]]in the case when the number of distinct configurations with given energy @xmath72 is exponentially large , @xmath73 will be finite as @xmath74 . due to the configurational entropy density , @xmath75 , the free energy density then differs from its minimum value , @xmath76 , even at the saddle point level . in mean field theory @xmath77
is finite below a temperature @xmath8 that coincides with the dynamic mode coupling temperature @xcite . at a lower temperature , @xmath78
, the entropy density vanishes like @xmath79and the system undergoes a transition into a statically frozen state . in order to have an explicit expression for @xmath75 , it is convenient to introduce the partition function@xcite @xmath80 , \]]leading to the free energy density @xmath81 , \label{m}\ ] ] such that @xmath82 gives the most probable value of @xmath72 which might also be written as @xmath83 .
\label{ftil0}\]]furthermore , the configurational entropy density follows from @xmath84 next we will describe the ways of calculating the configurational entropy density @xmath85 . a systematic approach to calculate @xmath86 was developed in ref .
the procedure of ref .
@xcite also reproduces the correct results for the configurational entropy in several random spin systems .
furthermore , it allows for a rather transparent motivation for the introduction of the variable @xmath87 of eq .
( [ m ] ) . in
what follows , we summarize the main idea and some technical steps of this formalism .
we consider the partition function in the presence of a bias configuration @xmath88:@xmath89 = \int d\varphi e^{-\beta \mathcal{h}\left [ \varphi \right ] -g\int d^{d}x\left ( \varphi \left ( \mathbf{x}\right ) -\psi \left ( \mathbf{x}\right ) \right ) ^{2}}.
\label{zpsi}\]]here , @xmath90 corresponds to the statistical sum over all density configurations of the system . at the end of our calculation
we will take the limit @xmath91 , however only after we take the thermodynamic limit .
for any finite @xmath11 , the configuration @xmath92 enters the problem as a quenched degree of freedom , analog to random fields in disordered spin systems .
we note that in distinction to the usual random field problem , the probability distribution function of metastable configurations @xmath93 $ ] is not expressed in terms of uncorrelated random numbers , but instead is determined by the partition function @xmath94 $ ] ( see below ) .
the free energy for a given bias configuration is @xmath95 = -t\log z\left [ \psi \right ] .\]]physically @xmath96 $ ] can be interpreted as the free energy for a metastable amorphous field configuration @xmath88 .
@xmath94 $ ] in eq .
( [ zpsi ] ) is obviously dominated by configurations @xmath97 that correspond to local minima of @xmath98 $ ] . in the replica formalism , no specific amorphous configuration need
be specified in order to perform the calculation .
rather , the assumption is made that the probability distribution for metastable configurations is determined by @xmath96 $ ] according @xmath99 \propto \exp \left ( -\beta _ { \mathrm{eff}}f\left [ \psi \right ] \right)\]]and is characterized by the effective temperature @xmath100 .
this allows one to determine the mean free energy @xmath101 p\left [ \psi \right]\]]and the corresponding mean configurational entropy @xmath102 \log p\left [ \psi \right ] . \label{sc}\ ] ] it is physically appealing then to introduce the free energy difference , @xmath103 , via @xmath104where @xmath103 gives the amount of energy lost if the system is trapped into locally stable states and hence not able to explore the entire phase space of the ideal thermodynamic equilibrium . if the limit @xmath91 behaves perturbatively , @xmath105
this indicates that the number of locally stable configurations stays finite in the thermodynamic limit , or at least grows less rapid than exponential with @xmath68 . in this case
all states are kinetically accessible . on the other hand ,
if the limit @xmath91 does not behave perturbatively , it means that the number of locally stable states , @xmath106,@xmath107 is exponentially large in @xmath68 .
this allows us to identify the difference between the equilibrium and typical free energy as an entropy : @xmath108the configurational entropy , @xmath109 , is extensive if there are exponentially many metastable states . within mean field theory , where the barriers are infinitely high
, the emergence of @xmath5 renders the system incapable of exploring the entire phase space .
@xmath4 is then the amount of entropy which the system that freezes it into a glassy state appears to lose due to its nonequilibrium - dynamics .
the approach of ref .
@xcite was successfully used to develop a mean field theory for glass formation in supercooled liquids mepa98.1,mepa98.2 yielding results in detailed agreement with earlier , non - replica approaches @xcite .
the mean values @xmath110 and @xmath111 can be determined from a replicated partition function @xmath112 = \int d\ \psi z^m\left [ \psi \right ] \label{zm}\]]via @xmath113 and @xmath114 with @xmath115and replica index @xmath116 .
inserting @xmath94 $ ] of eq .
( [ zpsi ] ) into eq .
( [ zm ] ) and integrating over @xmath117 one gets @xmath112 = \int d^{m}\varphi e^{-\beta \sum_{a=1}^{m}\mathcal{h}\left [ \varphi _ { a}\right ] + g\sum_{a , b=1}^{m}\int d^{d}x\varphi _ { a}\left ( x\right ) \varphi _ { b}\left ( x\right ) } \label{zmm}\]]which has a structure similar to a conventional equilibrium partition function .
the ergodicity breaking field @xmath118 causes a coupling between replicas which might spontaneously lead to order in replica space even as @xmath91 .
this spontaneous coupling between different copies of the system is then associated with a finite @xmath4 and thus glassiness . with @xmath119 , expressions for @xmath110 and @xmath4 follow from the replicated free energy @xmath120 via : @xmath121these results are in analogy to the usual thermodynamic relations between free energy ( @xmath122 ) , internal energy ( @xmath123 ) , entropy ( @xmath124 ) and temperature ( @xmath125 ) , see also ref .
if the liquid gets frozen in one of the many metastable states , the system can not anymore realize its configurational entropy , i.e. the mean free energy of frozen states is @xmath110 , higher by @xmath126 if compared to the equilibrium free energy of the liquid , fig .
[ fig4 ] . the physically intuitive analogy between effective temperature , mean energy , @xmath110 and configurational entropy to thermodynamic relations , eq .
( [ thermo ] ) , suggest that one should analyze the corresponding _ configurational heat capacity _ @xmath127using the distribution function eq .
( [ distrms ] ) we find , as expected , that @xmath128 is a measure of the fluctuations of the energy and configurational entropy of glassy states .
we obtain for the configurational heat capacity @xmath129where @xmath130 and @xmath131 . here
the mean values @xmath110 and @xmath111 are determined by eq .
( [ thermo ] ) .
the fluctuations of the configurational entropy and frozen state energy are then determined by@xmath132 \log ^{2}p\left [ \psi \right]\]]and@xmath133 f\left [ \psi \right ] ^{2},\ ] ] respectively .
both quantities can be expressed within the replica formalism in terms of a second derivative of @xmath134 with respect to @xmath87 .
for example it follows : @xmath135it is then easy to show that eq .
( [ ffluctc ] ) holds . with the introduction of the configurational heat capacity into the formalism we have a measure for the deviations of the number of metastable states from their mean value .
the analogy of these results to the usual fluctuation theory of thermodynamic variables @xcite further suggests that @xmath128 also determines fluctuations of the effective temperature with mean square deviation : @xmath136since @xmath128 is extensive , fluctuations of intensive variables , like @xmath137 , or densities , like @xmath138 , vanish for infinite systems .
however , they become relevant if one considers finite subsystems or small droplets . in the context of glasses
this aspect was first discussed in ref .
@xcite .
if the replica theory is marginally stable , i.e. the lowest eigenvalue of the fluctuation spectrum beyond mean field solution vanishes , it was shown in ref .
@xcite that @xmath137 agrees with the result obtained from the generalized fluctuation - dissipation theorem in the dynamic description of mean field glasses @xcite . typically , the assumption of marginality is appropriate for early times right after a rapid quench from high temperatures . in this case
@xmath139 for @xmath140 , i.e. the distribution function of the metastable states is not in equilibrium on the time scales where mode coupling theory or the requirement for marginal stability applies . above the kauzmann temperature it is however possible to consider the situation where @xmath141 , i.e. where the distribution of metastable states has equilibrated with the external heat bath of the system . since we are interested in the restoration of ergodicity for @xmath142 we use @xmath141 . below the kauzmann temperature
the assumption @xmath141 can not be made any longer as it implies a negative configurational entropy , inconsistent with the definition eq .
( [ sc ] ) .
it implies that the glass transition is now inevitable even if one could wait for arbitrarily long times .
while this is likely an idealization of mean field theory , it demonstrates that the ageing behavior below and above the kauzman temperature are very distinct . ultimately , glass formation is a dynamic phenomenon .
it is therefore useful and illustrative to offer a dynamical interpretation of the replica formalism of ref .
below we will provide a qualitative set of arguments .
a detailed and quantitative derivation of the formalism , based on the dynamic theory of glassy systems by kurchan and cugliandolo @xcite , is given in ref .
@xcite .
consider a system characterized by a field variable @xmath143 .
let us assume that certain field configurations evolve extremely slowly , and equilibrate at a much later times than others . in terms of particle coordinates , a natural choice for slow and fast variables would be the fiducial position @xmath144 and the harmonic vibrations @xmath145 , according to @xmath146 @xmath147 . in our field theoretic description we assume that slow field configurations are characterized by @xmath148 while we continue to refer to fast configurations as @xmath143 .
let the dynamics of the fast degrees of freedom be governed by the langevin equation @xmath149where @xmath150 \
+ \frac{g}{2}\int dx\left ( \varphi \left ( x\right ) -\psi \left ( x\right ) \right ) ^{2}\]]contains a weak coupling ( @xmath151 ) between fast and slow degrees of freedom , just like in the replica approach discussed above .
fast variables @xmath152 are subject to white noise with temperature @xmath153 @xmath154the equilibrium distribution function for fixed @xmath118 is then @xmath155 = \frac{1}{z\left [ \psi \right ] } \exp \left ( -\beta \mathcal{s}\left ( \varphi , \psi \right ) \right)\]]where @xmath94 = \int d\varphi \exp \left ( -\beta \mathcal{s}\left [ \varphi , \psi \right ] \right ) $ ] .
next we assume that the dynamics of the slow degrees of freedom is then governed by @xmath156 } { \delta \psi \left ( x , t\right ) } + \eta _ { s}\left ( x , t\right ) .\]]with
@xmath96 = -t\log z\left [ \psi \right ] $ ] .
thus , during the entire dynamics of the slow degrees of freedom , the fast modes are assumed to be in equilibrium already .
it is important to assume different noise for the slow variables @xmath157 and for the fast variables @xmath158 .
the variables @xmath148 will , in general , not equilibrate with temperature @xmath153 , but may be characterized by an effective temperature @xmath159 , i.e. we write : @xmath160then , the stationary probability distribution for @xmath118 is @xmath161 = \frac{1}{z}\exp \left ( -\beta _ { \mathrm{eff}}f\left [ \psi \right ] \right ) = \frac{z\left [ \psi \right ] ^{m}}{z}\]]with @xmath162 . the joint distribution of @xmath163 and @xmath118 can now be written as @xmath164 = p\left [ \psi \right ] p\left [ \varphi |\psi \right ] $ ] .
it is straightforward to show that this distribution gives results identical to the replica approach . above the kauzmann temperature ,
@xmath165 , equilibration of the slow variables is still possible and @xmath153 approaches @xmath137 , i.e. @xmath166
to gain detailed insight into the possibility of glass formation , we now go back to our model described by and analyze the problem ( [ hint ] ) with @xmath167 analytically within the self consistent screening approximation ( scsa ) @xcite , an approximate treatment that is controlled by an expansion in @xmath168 , where @xmath58 is the number of components of @xmath163 , see ref . @xcite .
the free energy @xmath169 of the replicated hamiltonian is given in terms of the regular correlation function @xmath170 and the correlation function @xmath171 for @xmath172 , i.e. between the fields in different replicas .
the latter corresponds to the edward - anderson parameter signaling a glassy state . for a system in the universality class of the random first order transition
we expect that below a temperature @xmath173 the system establishes an exponentially large number of metastable states and long time correlations , characterized by the correlation function @xmath174 .
these long time correlations occur even though no state with actual long range spatial order exists . within the scsa ,
the relevant part of the free energy @xmath169 is : @xmath175which determines the configurational entropy @xmath176 . here ,
@xmath177is the correlation function matrix with @xmath178 and @xmath179 .
the symbol in eq .
( [ fe ] ) includes the trace of the replica space and the momentum integration .
the matrix @xmath180 is related to @xmath181 via @xmath182where @xmath183is the generalized polarization matrix .
the symbol @xmath184 denotes a convolution in fourier space .
the replicated schwinger - dyson equation can be written as@xmath185 where @xmath186 is the self - energy matrix and @xmath187 the hartree correlation function of eq .
( [ braz ] ) . within the scsa
the self - energy has diagonal elements @xmath188 and off diagonal elements @xmath189 in replica space , where @xmath190 and @xmath191 being , respectively the diagonal and off - diagonal elements of @xmath180 .
these equations form a closed set of self - consistent equations which enable us to solve for @xmath192 and @xmath193 , and then determine the configurational entropy . to evaluate @xmath169
we use the fact that a matrix of the form @xmath194 has @xmath195 degenerate eigenvalues @xmath196 where @xmath197 and and one eigenvalue @xmath198 .
this yields @xmath199 this result allows us to evaluate @xmath134 of eq .
( [ fe ] ) . performing the derivative of the resulting expression with respect to @xmath87
then yields the configurational entropy @xmath200 -s\left [ \frac{f\otimes f}{(u_{4}t)^{-1}+g\otimes g}\right ] \right\ } , \label{sconf}\]]were , @xmath201\equiv -x-\ln ( 1-x)$ ] .
the analysis of these coupled equations reveals that the self energies @xmath202 and @xmath203 are only weakly momentum dependent .
then the impact of @xmath202 is solely to renormalize the correlation length @xmath204 .
however , the emergence of @xmath203 leads to a qualitative change .
dimensional analysis reveals that @xmath203 is an inverse length squared , which motivates one to introduce the lindemann length @xmath205 of the glass via @xmath206 .
an interpretation of this length scale in terms of slow defect motion in a stripe glasses was given in ref .
@xcite . by inspection of the dyson equation and using
the fact that @xmath207 is strongly peaked at the modulation wave number @xmath53 one finds for @xmath208 that @xmath209@xmath207 vanishes rapidly away from the peak ( as does @xmath210 ) and it follows from the same equation , [ dyson ] , that for large @xmath211 holds:@xmath107 @xmath212if a solution for @xmath193 exists , it is going to be a peaked at @xmath53 , but smaller and narrower than @xmath192 .
consequently , if a stripe glass occurs , the long time limit of the correlation function is not just a slightly rescaled version of the instantaneous correlation function , but it is multiplied by a @xmath213 dependent function that leads to a qualitatively different behavior for different momenta
. once a glassy state is formed , configurations which contribute to the peaks of @xmath207 and @xmath214 , i.e. almost perfect stripe configurations , are almost unchanged even after long times .
close to @xmath53 , @xmath215 is solely reduced by some momentum independent debye - waller factor @xmath216 .
on the other hand , configurations which form the tails of @xmath207 , i.e. defects and imperfections of the stripe pattern , disappear after a long time since now @xmath217 .
the ratio of both functions is now strongly momentum dependent .
@xmath218 becomes sharper than @xmath207 because local defects got healed in time .
the length scale that determines the transition between these two regimes is the length @xmath205 .
this length can therefore be associated with the allowed generalized vibrational motions in a potential minima of the complex energy landscape of the system . in analogy with structural glasses we therefore call @xmath205 the lindemann length of the stripe glass .
is an average distance over which the defects are manage to wander ( lindemann length ) and @xmath219 is the correlation length ( see text ) .
upper panel illustrates the situation when defects are relatively close to each other so that they disappear with time and system effectively heals itself .
bottom panel depict the opposite situation when wandering of the defects does not produce healing since they are on distances @xmath220 from each other ( see ref .
sww00 for details . ) ] for our subsequent analysis we introduce the dimensionless quantity @xmath221 via @xmath222i.e .
we are now in a position to perform the momentum integration and evaluate the configurational entropy as function of @xmath221 .
it follows that @xmath224 \label{sofq}\]]where@xmath225 in fig .
[ fig6 ] we show our results for @xmath226 for different values of @xmath227 .
a self consistent solution of the above set of coupled equations corresponds to finding stationary points @xmath228 with @xmath229@xmath230 is always a solution with @xmath231 .
in addition there is a locally stable solution for @xmath232 with configurational entropy given as @xmath233 and @xmath234 .
the configurational entropy at the metastable maximum vanishes at @xmath235 with @xmath236 .
this is the behavior that is generally expected in a system a with random first order transition . upon lowering the temperature the ergodic ,
liquid like state becomes more and more correlated , i.e. the correlation length growths .
once @xmath204 reaches a threshold value @xmath237 about twice the typical length scale for field modulations @xmath238 , glassy dynamics sets in as consequence of an emergence of exponentially many states .
the kauzman temperature is reached once the liquid state correlation length grows even further and becomes equal to @xmath239 . as a function of the dimensionless parameter @xmath240 , eq .
( 4.10 ) for different values of another parameter @xmath241 .
it is instructive to compare this plot with the schematic plot of fig .
it is also interesting to analyze the penalty for a spatially varying overlap @xmath242 .
this will be essential for the landau expansion and the droplet arguments that follow below . performing a gradient expansion of the replica free energy yields the correction @xmath243 where @xmath244\ ] ] the prefactor @xmath245 appears from our definition for @xmath221 ( [ sigmaf ] ) . for @xmath30 and @xmath246 the integration can be performed exactly with the following result @xcite : @xmath247 as was anticipated in ref .
@xcite , the penalty for a spatial deformation becomes small as @xmath248 , i.e. in the limit where the modulation length is very large .
note that at @xmath249 or @xmath250 , the dimensionless product @xmath251 is independent of @xmath252 , i.e. varying the characteristic scale yields @xmath253 .
thus , we analyzed an analytically treatable example for a random first order transition which enabled us to identify the underlying physical mechanism for glassiness in a uniformly frustrated system .
once equilibrium correlations are sufficiently strong and the stripe liquid - stripe solid transition is kinematically inaccessible , the system is governed by slow glassy dynamics . at this point
a lindemann length , @xmath205 , emerges , which is a length scale over which imperfections of the stripe pattern manage to wander , fig .
defects of the perfectly ordered state are still abundant as the healing length is finite , i.e. the system enters a random solid state .
local order is however established .
this may not be identical to the most stable crystalline , i.e. long ranged ordered , configuration , but rather correspond to configurations that are occurring with high weight close to the spinodal of the system . in the context of supercooled liquids , the natural analog of such locally correlated configuration are icosahedral or body centered cubic short range order .
computer simulations on the model discussed above with @xmath57 , suggest that the nucleation barriers of the crystalline state are rather low geissler04 .
however , the inclusion of asymmetric terms in the interaction potential ( @xmath254 ) will likely strengthen first order transitions to a crystalline state and enhance the nucleation barriers .
this behavior was analyzed in refs .
the analysis of the previous section demonstrated that uniformly frustrated systems undergo self generated glass transitions with an entropy crisis , of the random first order transition universality class .
these calculations are on the mean field level and ignore issues related to dynamical heterogeneity , i.e. variations in the typical time scales in spatially nearby regions .
the relevance of such effect is already evident from the fact that fluctuations in the heat capacity become important in finite regions , as shown above . to capture these phenomena we need to go beyond the mean field level and include droplet formations .
droplets are not regions with different values of the field @xmath255 that underlies our analysis . instead , by droplets we mean a spatial variation of the lindemann length , i.e. of the parameter @xmath221 used above to characterize the overlap between configurations at distant times .
thus , it is useful to consider instead of @xmath221 a dynamic field @xmath256 that generates the replica partition function , i.e. @xmath257 \right ) .\]]the mean field solution of this problem is then determined by @xmath258 /\delta
q_{ab}=0 $ ] with the ansatz @xmath259with @xmath228 playing the exact same role as the corresponding quantity determined by the condition eq .
( [ sstationary ] ) in the previous section .
formally @xmath260 $ ] can be obtained from the field theory discussed above by multiplying @xmath261 of eq .
( [ zmm ] ) by unity@xmath262 \right\ } \end{split}\]]and switch the order of integration .
then we find @xmath263 } = \int d\mu d^{m}\varphi e^{-\sum_{a=1}^{m}\beta \mathcal{h}\left [ \varphi ^{a}\right ] } \times \\ & \times e^{-\sum_{a , b=1}^{m}\int_{\mathbf{x , x}^{\prime } } q_{ab}(\mathbf{x , x}^{\prime } ) \left ( \mu _ { ab}(\mathbf{x , x}^{\prime } ) -\varphi ^{a}(\mathbf{x})\varphi ^{b}(\mathbf{x}^{\prime } ) \right ) } .
\end{split}\]]we consider the fourier transform of @xmath264 with respect to the relative coordinates @xmath265 and approximate it by its value at @xmath266 .
this is consistent with our finding that the self energies of the mean field theory depend only weakly on momentum .
we do however include the dependence on the center of gravity coordinate to analyze spatially varying overlaps between distant configurations , yielding @xmath267 . in order to keep our calculation transparent
we will not analyze the model discussed for uniformly frustrated systems , but instead start from a simpler landau theory that is in the same universality class @xcite : @xmath268 = e_{0}\sum_{a , b}\int \frac{d^{3}r}{a_{0}^{3}}\left ( h\left [ q_{ab}\right ] -\frac{u}{3}\sum_{c}q_{ab}q_{bc}q_{ca}\ \right ) \label{ham1}\]]with @xmath269 = \frac{a_{0}^{2}}{2}\left ( \nabla q_{ab}\right ) ^{2}+\frac{t}{2}q_{ab}^{2}-\frac{u+w}{3}q_{ab}^{3}+\frac{y}{4}q_{ab}^{4 } \label{ham2}\]]and replica index @xmath270,@xmath271 .
@xmath272 is a length scale of the order of the first peak in the radial distribution function of the liquid and @xmath273 is a typical energy of the problem that determines the absolute value of @xmath274 .
in addition the problem is determined by the dimensionless variables @xmath275 , @xmath276 , @xmath277 and @xmath278 , which are in principle all temperature dependent .
we assume that the primary @xmath153-dependence is that of the quadratic term , where @xmath279 . in what follows we
further simplify the notation and measure all energies in units of @xmath273 and all length scales in units of @xmath280 .
formally , eqs .
( [ ham1 ] ) and ( [ ham2 ] ) can be motivated as the taylor expansion of the free energy @xmath281 $ ] with respect to @xmath282 . in practice
, the explicit taylor expansions of more microscopic models do not yield quantitatively reliable results .
for example , the taylor expansion of eq .
( [ sofq ] ) poorly reproduces the behavior in the glassy regime .
still , the simple landau expansion of eqs .
( [ ham1 ] ) and ( [ ham2 ] ) allows us to gain key insights that enable us to investigate the more general case .
in particular , we will demonstrate that the replica instanton theory can easily be generalized to the more complex stripe glass model .
the mean field analysis of the model hamiltonian eqs .
( [ ham1 ] ) and ( ham2 ) is straightforward . inserting the ansatz eq .
( [ mf ] ) into @xmath260 $ ] and minimizing w.r.t @xmath221 yields @xmath230 or @xmath283nontrivial solutions exits for @xmath284 with @xmath285 , which determines the mode coupling temperature @xmath286 . inserting @xmath228 of eq .
( [ qstar ] ) into @xmath260 $ ] yields at the saddle point @xmath287 $ ] for the replica free energy , which yields : @xmath288 the configurational entropy , as determined by @xmath289 is given by @xmath290the stationary points of @xmath226 are given by @xmath228 of eq.[qstar ] .
inserting @xmath228 yields that @xmath4 vanishes at @xmath291 with @xmath292 .
close to @xmath293 it follows that @xmath294as expected . at @xmath295 one
finds @xmath296 .
@xmath297for the configurational heat capacity follows@xmath298it holds at @xmath299 and we can write@xmath300thus , we see that the main findings of the model calculations are reproduced by the simple landau expansion , eqs .
( [ ham1 ] ) and ( [ ham2 ] ) .
they will now be used as starting point for our analysis of dynamical heterogeneity in form of a replica instanton theory .
at the mean field level a glass at @xmath140 is frozen in one of many metastable states .
@xmath228 then characterizes the overlap between configurations at distant times .
the free energy of such a frozen state is higher by @xmath126 compared to the ergodic liquid state that is characterized by @xmath230 .
thus , for @xmath142 the mean field glass is locally stable .
local stability also follows from the fact that the lowest eigenvalue of the fluctuation matrix @xmath301 is positive for @xmath302 and vanishes at @xmath10 , see ref .
the @xmath221-dependence of @xmath226 shown in fig .
[ fig6 ] suggests that the decay modes for the frozen state are droplet excitations , similar to the nucleation of an unstable phase close to a first order transition .
this situation was analyzed in ref .
franz05 - 1,dzero05 . in agreement with the rfot theory @xcite ,
the driving force for nucleation is the configurational entropy , leading to the notion of entropic droplets .
the formal approach to analyze entropic droplets is performed in terms of the effective potential approach of refs .
we used this technique to formulate a replica instanton and barrier fluctuation theory in refs .
dzero05,dzero09 . in what follows we use a slightly simpler approach that yields essentially the same results , but is physically significantly more transparent . as can be motivated by the more general effective potential approach of refs .
@xcite , instanton solutions for entropic droplets can be determined from@xmath303 } { \delta q_{ab}\left ( \mathbf{x}\right ) } = 0 , \label{inststat}\]]where we allow for spatial variations of the overlap @xmath304 .
this yields the following nonlinear equation for @xmath305 : @xmath306with configurational entropy density @xmath307 . in case of the landau expansion we use @xmath226 of eq .
( [ scmon ] ) , while for the stripe glass approach we start from eq .
( [ sofq ] ) .
we will first analyze the simpler case of the landau theory .
( [ inst1 ] ) admits an exact solution in the thin wall limit @xmath308 : @xmath309 , \label{pcthinwall}\]]where the integration constant @xmath310 is a function of @xmath311 and @xmath278 .
@xmath312 is the droplet radius and @xmath204 is the interface width given by @xmath313inserting the solution eq .
( [ pcthinwall ] ) into the expression into @xmath314 $ ] we calculate the value of the mean barrier .
the latter is determined by optimizing the energy gain due to creation of a droplet and energy loss due to the surface formation . as a result for the mean barrier we find ( reintroducing the energy scale @xmath273 and length scale @xmath280 ) @xmath315the droplet radius @xmath316is determined from the balance between the interface tension and the entropic driving force for nucleation .
furthermore , @xmath317 is the order parameter at the kauzmann temperature .
when temperature approaches the @xmath165 the radius of the droplet as well as the mean barrier diverge .
one finds @xmath318 and @xmath319 .
since the droplet interface @xmath204 remains finite as @xmath320 , the thin wall approximation is well justified close to the kauzman temperature . on the other hand , @xmath312 and @xmath204
become comparable for temperatures close to @xmath8 and the thin wall approximation breaks down . combining @xmath321 and @xmath322 , we obtain @xmath323 for the exponent that relates the droplet size @xmath312 and the configurational entropy density : @xmath324 .
close to @xmath293 follows @xmath325 with @xmath326 the critical droplet nucleation radius is @xmath327 yielding the barrier @xmath328for the nucleation of entropic droplets .
this leads to a mean relaxation time of @xmath329 it was pointed out in ref .
@xcite that wetting effects of the interface alter the a relationship between droplet size and entropic driving force to @xmath330 with exponent @xmath331 .
a renormalization of the droplet interface due to wetting of intermediate states on the droplet surface was shown to yield @xmath332 @xcite , leading to @xmath333 and correspondingly to a vogel - fulcher law @xmath334 for the mean relaxation time@xmath335 it is straightforward to analyze the more complex configurational entropy of the stripe glass problem . in this case
the nonlinear instanton equation can not be solved analytically .
however we can make progress by performing a variational calculation for the droplet .
the mean barrier is @xmath336 -s_{c}\left ( q^{\ast } \right ) , \]]where @xmath337 is a localized instanton solution which differs from @xmath228 in a finite region .
we make the trial ansatz : @xmath338 and insert it into the above expression for @xmath339 $ ] .
we find @xmath340with surface tension @xmath341\]]minimizing with respect to the droplet wall thickness @xmath342 yields@xmath343and correspondingly for the surface tension @xmath344using the our result eq .
( [ sofq ] ) for the configurational entropy as a function of @xmath221 for the stripe glass problem in @xmath30 and for @xmath345 ( i.e. with the long range coulomb interaction @xmath346 ) and eq .
( [ a02 ] ) for the gradient coefficient @xmath280 , yields at @xmath165 : @xmath347where @xmath348 .
thus we see that the surface tension of entropic droplets vanishes in the limit @xmath349 . for the wall thickness
we obtain @xmath350 .
finally , we use our results for the surface tension to analyze the variation of the fragility @xmath351 in eq .
( [ taufrag ] ) as function of @xmath252 .
one finds @xmath352 which was further investigated in a numerical analysis of the stripe glass problem in ref .
this analysis led to @xmath353 .
if we make the following estimate @xmath354 and neglect the dependence of @xmath355 on @xmath252 , we find that @xmath356 in qualitative agreement with the results of ref .
@xcite .
numerous experiments on supercooled liquids are not only sensitive to the mean barrier , @xmath357 , but are able to measure the entire ( broad ) excitation spectrum in glasses @xcite .
most notably , the broad peaks in the imaginary part of the dielectric function @xmath358 are most naturally understood in terms of a distribution @xmath359 of relaxation times , such that @xmath360similarly dynamical heterogeneity with spatially fluctuating relaxation times yields non - exponential ( frequently stretched exponential ) relaxation of the correlation function @xmath361other effects that are most likely caused by a distribution of relaxation rates include the break down of the stokes - einstein relation @xmath362 between the diffusion coefficient @xmath351 of a particle of size @xmath363 and the viscosity @xmath364 @xcite .
these experiments call for a more detailed analysis of the fluctuations @xmath365of the activation barriers and , more generally , of the distribution function @xmath366 of barriers . the latter yields
the distribution function of the relaxation times @xmath367through @xmath368 .
for example , in case of a gaussian distribution of barriers one obtains a broad , log - normal distribution of relaxation rates:@xmath369with @xmath370and @xmath371 from eq .
( [ meanr ] ) . while the distribution , eq .
( [ lognorm ] ) , does not yield a stretched exponential form for the correlation function , it can often be approximated by @xmath372with @xmath373 .
furthermore , the study of higher order moments of @xmath366 is important to determine whether the distribution is indeed gaussian or more complicated . in ref .
@xcite we used the replica formalism discussed here as well as the more elaborate replica method of formalism refs .
franz95,franz97,barrat97 to determine higher order moments of the barrier distribution function . using the `` thin wall '' approximation for @xmath374 given by eq .
( [ pcthinwall ] ) we obtain for the second moment @xmath375where @xmath312 is the radius of the droplet where the explicit expressions for the coefficient @xmath376 and length @xmath377 are given in ref .
it is noteworthy that there is a surface contribution to the moment of the barrier fluctuations that is a consequence of correlations between droplet and homogeneous background with overlap @xmath228 . in ref .
@xcite it was also shown that the barrier distribution function @xmath366 is gaussian , at least if one considers the first six moments .
the skewness of the actual barrier distribution measured in structural glasses is , in our view , an effect due to the interaction of spatially overlapping instantons .
to summarize , we have discussed several models of glassy systems where the randomness is self generated rather than induced by strong external factors .
in particular , we applied the replica formalism developed for the spin glass systems to study the glass transition in the many - body systems at the presence of an arbitrary weak disorder .
we employed the landau theory to analyze the mean field glass transition using the saddle point approximation .
we have also considered the energy fluctuations around the saddle point and evaluated the barrier height distribution .
we are grateful to harry westfahl jr . for discussions and collaborations on problems discussed in this chapter .
this research was supported by the ames laboratory , operated for the u.s .
department of energy by iowa state university under contract no .
de - ac02 - 07ch11358 ( j. s. ) , a fellowship of the institute for complex adaptive matter , by the intelligence advanced research projects activity ( iarpa ) through the us army research office award w911nf-09 - 1 - 0351 and kent state university ( m.d . ) , and the national science foundation grant che-0317017 ( p. g. w. ) .
d. r. nelson , phys .
50 982 ( 1983 ) ; d. r. nelson , phys .
b * 28 * 5515 ( 1983 ) ; s. sachdev and d. r. nelson , phys .
. lett . * 53 * 1947 ( 1984 ) ; s. sachdev and d. r. nelson , phys .
b * 32 * 1480 ( 1985 ) . | we review several models of glassy systems where the randomness is self generated , i.e. already an infinitesimal amount of disorder is sufficient to cause a transition to a non - ergodic , glassy state .
we discuss the application of the replica formalism developed for the spin glass systems to study the glass transition in uniformly frustrated many - body systems . here
a localization in configuration space emerges leading to an entropy crisis of the system . using a combination of density functional theory and landau theory of the glassy state ,
we first analyze the mean field glass transition within the saddle point approximation .
we go beyond the saddle point approximation by considering the energy fluctuations around the saddle point and evaluate the barrier height distribution . |
the evolution of the cellular network generations is influenced primarily by continuous growth in wireless user devices , data usage , and the need for a better quality of experience ( qoe ) .
more than 50 billion connected devices are expected to utilize the cellular network services by the end of the year 2020 @xcite , which would result in a tremendous increase in data traffic , as compared to the year 2014 @xcite
. however , state - of - the - art solutions are not sufficient for the challenges mentioned above . in short , the increase of 3d ( device ,
data , and data transfer rate ) encourages the development of 5 g networks .
specifically , the fifth generation ( 5 g ) of the cellular networks will highlight and address the following three broad views : ( _ i _ ) user - centric ( by providing 24@xmath07 device connectivity , uninterrupted communication services , and a smooth consumer experience ) , ( _ ii _ ) service - provider - centric ( by providing a connected intelligent transportation systems , road - side service units , sensors , and mission critical monitoring / tracking services ) , and ( _ iii _ ) network - operator - centric ( by providing an energy - efficient , scalable , low - cost , uniformly - monitored , programmable , and secure communication infrastructure ) .
therefore , 5 g networks are perceived to realize the three main features as below : * _ ubiquitous connectivity _ : in future , many types of devices will connect ubiquitously and provide an uninterrupted user experience .
in fact , the user - centric view will be realized by ubiquitous connectivity .
* _ zero latency _ : the 5 g networks will support life - critical systems , real - time applications , and services with zero delay tolerance .
hence , it is envisioned that 5 g networks will realize zero latency , _
i_._e_. , extremely low latency of the order of 1 millisecond @xcite .
in fact , the service - provider - centric view will be realized by the zero latency . * _ high - speed gigabit connection _ : the zero latency property could be achieved using a high - speed connection for fast data transmission and reception , which will be of the order of gigabits per second to users and machines @xcite .
a few more _ key features of 5 g networks _ are enlisted and compared to the fourth generation ( 4 g ) of the cellular networks , as below @xcite : ( _ i _ ) 10-@xmath1 number of connected devices , ( _ ii _ ) 1000@xmath2 higher mobile data volume per area , ( _ iii _ ) 10-@xmath1 higher data rate , ( _ iv _ ) 1 millisecond latency , ( _ v _ ) 99.99% availability , ( _ vi _ ) 100% coverage , ( _ vii _ ) @xmath3 energy consumption as compared to the year 2010 , ( _ viii _ ) real - time information processing and transmission , ( _ ix _ ) @xmath4 network management operation expenses , and ( _ x _ ) seamless integration of the current wireless technologies .
r9 cm [ fig : circle ] the revolutionary scope and the consequent advantages of the envisioned 5 g networks , therefore , demand new architectures , methodologies , and technologies ( see figure [ fig : circle ] ) , _ e_._g_. , energy - efficient heterogeneous frameworks , cloud - based communication ( software - defined networks ( sdn ) and network function virtualization ( nfv ) ) , full duplex radio , self - interference cancellation ( sic ) , device - to - device ( d2d ) communications , machine - to - machine ( m2 m ) communications , access protocols , cheap devices , cognitive networks ( for accessing licensed , unlicensed , and shared frequency bands ) , dense - deployment , security - privacy protocols for communication and data transfer , backhaul connections , massive multiple - input and multiple - output ( mmimo ) , multi - radio access technology ( rat ) architectures , and technologies for working on millimeter wave ( mmwave ) 30300 ghz .
interestingly , the 5 g networks will not be a mere enhancement of 4 g networks in terms of additional capacity ; they will encompass a system architecture visualization , conceptualization , and redesigning at every communication layer @xcite .
several industries , _ e_._g_. , alcatel - lucent @xcite , docomo @xcite , gsma intelligence @xcite , huawei @xcite , nokia siemens networks @xcite , qualcomm @xcite , samsung @xcite , vodafone , the european commission supported 5 g infrastructure public private partnership ( 5gppp ) @xcite , and mobile and wireless communications enablers for the twenty - twenty information society ( metis ) @xcite , are brainstorming with the development of 5 g networks .
currently , the industry standards are yet to be evolved about the expected designs and architectures for 5 g networks .
* scope of the paper .
* in this paper , we will review the vision of the 5 g networks , advantages , applications , proposed architectures , implementation issues , real demonstrations , and testbeds .
the outline of the paper is provided in figure [ fig : outline of the paper ] . in section [
section : promises of 5 g ] , we will elaborate the vision of 5 g networks .
section [ section : challenges in the development of 5 g ] presents challenges in the development of 5 g networks .
section [ sec : architectures of the future 5 g mobile cellular networks ] addresses the currently proposed architectures for 5 g networks , _
e_._g_. , multi - tier , cognitive radio based , cloud - based , device proximity based , and energy - efficient architectures . section [ sec : management issues in 5 g networks ] presents issues regarding interference , handoff , quality of services , load balancing , channel access , and security - privacy of the network .
sections [ sec : methodology and technology for 5 g networks ] , [ sec : applications of 5 g networks ] , and [ sec : real demonstrations and test - beds for 5 g networks ] present several methodologies and technologies involved in 5 g networks , applications of 5 g networks , and real demonstrations and testbeds of 5 g networks , respectively .
we would like to emphasize that there do exist some review works on 5 g networks by andrews et al .
@xcite , chvez - santiago et al .
@xcite , and gavrilovska et al .
@xcite , to the best of our knowledge . however , our perspective about 5 g networks is different , as we deal with a variety of architectures and discuss several implementation affairs , technologies in 5 g networks along with applications and real - testbed demonstrations .
in addition , we intentionally avoid an mmwave oriented discussion in this paper , unlike the current work @xcite .
we encourage our readers to see an overview about the generations of the cellular networks ( see table [ table : the generation of cellular networks ] ) and the crucial limitations of current cellular networks in the next section .
.the generations of the cellular networks . [ cols="<,<,<,<",options="header " , ] wang et al .
@xcite suggested a way for separating indoor and outdoor users and using a _ mobile small - cell _ on a train or a bus . for separating indoor and outdoor users
, a mbs holds large antenna arrays with some antenna elements distributed around the macrocell and connected to the mbs using optical fibers . a sbs and large antenna arrays are deployed in each building for communicating with the mbs .
all ues inside a building can have a connection to another ue either through the sbs or by using wifi , mmwave , or vlc .
thus , the separation of users results in less load on a mbs .
wang et al .
@xcite also suggested to use a mobile small - cell that is located inside a vehicle to allow communication among internal ues , while large antenna arrays are located outside the vehicle to communicate with a mbs .
thus , all the ues inside a vehicle ( or a building ) appear to be a single unit with respect to the corresponding mbs , and clearly , the sbs appears as a mbs to all these ues . in @xcite ,
a two - tier architecture is deployed as a process of _ network densification _ that is a combination of _ spatial densification _ ( increasing the number of antennas per ue and mbs , and increasing the density of bss ) and _ spectral aggregation _ ( using higher frequency bands @xmath5 3 ghz ) .
a tradeoff between the transmission power of a macrocell and the coverage area of small - cells is presented in @xcite ; _
i_._e_. , on the one hand , if the transmission power of a macrocell is high , then many adjacent ues to a small - cell may find themselves in the service area of the macrocell , and hence , it will decrease the coverage area of that small - cell . on the other hand , if the transmission power of a macrocell is low , then the coverage area of the small - cell will increase .
therefore , _ cell range expansion _
( _ i_._e_. , a biased handoff in the favor of small - cells ) is carried out to serve more ues by small - cells to which they are closer .
moreover , sbss , deployed in offices or homes , can be used to serve outdoor users , _
e_._g_. , pedestrians and low - mobility vehicles , in their neighborhoods , and the approach is called _ indoor - to - outdoor user service _ @xcite .
hossain et al .
@xcite presented a multi - tier architecture consisting of several types of small - cells , relays , and d2d communication for serving users with different qos requirements in spectrum - efficient and energy - efficient manners .
interestingly , all of these architectures consider that ues spontaneously discover a sbs .
zhang et al .
@xcite proposed a centralized system in which a mbs assists ues to have connections to particular sbss , thereby interference between ues and sbss is reduced . however , this approach overburdens the mbs .
* advantage of the deployment of small - cells . * * _ high data rate and efficient spectrum use _ : the small physical separation between a sbs and ues ( served by the same sbs ) leads to a higher data rate and a better indoor coverage . also , the spectrum efficiency increases due to fewer ues in direct communication with a mbs @xcite . *
_ energy saving _ :
the use of small - cells reduces the energy consumption of the network ( by not involving mbss ) and of ues ( by allowing ues to communicate at a shorter range with low signaling overhead ) @xcite . *
_ money saving _ : it is more economical to install a sbs without any cumbersome planning as compared to a mbs , and also the operational - management cost is much lower than the cost associated with a mbs @xcite .
* the plug - and - play utility of small - cells boosts the on - demand network capacity @xcite . * _ less congestion to a mbs _ : sbss offload ues from a mbs so that the mbs is lightly loaded and less congested , and hence , improve the system capacity @xcite . * _ easy handoff _ : mobile small - cells also follow the advantages of small - cells . moreover , they provide an attractive solution to highly mobile ues by reducing handoff time overheads , since a mobile small - cell is capable to do the handoff on behalf of all related ues @xcite .
* disadvantage of small - cells . * despite numerous prominent benefits as mentioned above , there are a few realistic issues such as implementation cost and operational reliability .
the small - cells indeed impose an initial cost to the infrastructure , but less than the cost associated with a mbs .
moreover , a frequent authentication is mandatory due to frequent handoff operations .
further , an active or passive ( on / off ) state update of any small - cell would definitely result in frequent topological updates . * open issues in the deployment of 2-tier architectures using small - cells . * * _ interference management _ : the deployment of small - cells results in several types of interferences , as : _ inter - tier interference _ ( _ i_._e_. , interference from a mbs to a sbs , interference from a mbs to a sbs s ues , and interference from a sbs to a mbs s ues ) , and _ intra - tier interference _ ( _ i_._e_. , interference from a sbs to other sbss ues ) .
hence , it is also required to develop models and algorithms to handle these interferences @xcite . *
_ backhaul data transfer _ : though we have models to transfer data from a sbs to the core network , which we will discuss next in section [ subsubsec : backhaul data transfer from small - cells ] , an extremely dense - deployment of small - cells requires a huge amount of data transfer , and certainly , requires cost efficient architectures .
data transfer from a sbs to the core network is a challenging task , and in general , there may be three approaches to transfer ( backhaul ) data to the core network , as follows : * _ wired optical fiber _ : by establishing a wired optical fiber link from each sbs to a mbs ; however , it is time - consuming and expensive .
* _ wireless point - to - multipoint _ ( ptmp ) : by deploying a ptmp - bs at a mbs that communicates with sbss and transfers data to the core network
. * _ wireless point - to - point _ ( ptp ) : by using directional antennas in line - of - sight ( los ) environments ; hence , it provides high capacity ptp links ( same as with wired optical fibers ) , at a significantly lower cost .
ge et al .
@xcite presented two architectures based on the wireless ptmp approach . in the first ( centralized ) architecture ,
all sbss send data using mmwave to a mbs that eventually aggregates the received data and forwards the same to the core network using fiber . in the second ( distributed ) architecture , all small - cells cooperatively forward data using mmwave to a specified sbs that transfers data to the core network using fiber without the explicit involvement of a mbs .
ni et al .
@xcite proposed an _ adaptive backhaul architecture _ based on the wireless ptp approach and frequency division duplex for ul and dl channels .
a tree structure is used , where the root node is connected to the core network using fiber , the leaf nodes represent ues , and other nodes represent sbss .
the data is transferred from the leaf nodes to the root node that transfers the same to the core network .
the bandwidth is selected dynamically for backhaul links , as per the current bandwidth requirements , interference conditions , and the network topology .
a similar approach is also presented in @xcite .
an automatic detection and recovery of a failed cell is an important issue in densely deployed multi - tier architectures .
wang and zhang @xcite provided three approaches for designing a self - healing architecture such as below : 1 .
_ centralized approach _ : a dedicated server is responsible for detecting a failed cell by measuring and analyzing abnormal behavior of users , _
e_._g_. , received signal strengths ( rsss ) at users and handoff by several users at any time from a particular cell .
the server collects global information and reconfigures the failed cell .
however , the approach suffers with a high communication overhead and a high computational cost .
2 . _ distributed approach _ :
each sbs detects failed small - cells in neighborhoods by measuring and analyzing users handoff behavior and the neighboring small - cells signals .
consequently , on detecting a failed cell , a sbs might increase the transmission power in order to incorporate users of the failed cell .
however , the approach might not work efficiently in case that users are scattered sparsely .
3 . _ local cooperative or hybrid approach _ : this approach combines the benefits of both the previous approaches , and therefore , minimizes the drawback . essentially , two steps are utilized , namely distributed trigger and cooperative detection . in the distributed trigger ,
each sbs collects information about users behavior .
subsequently , a trigger message is sent to a dedicated server in case the received information thrives below a certain threshold .
hence , it does not require communication among small - cells . in the cooperative detection
, the dedicated server takes the final decision based on the information received from several small - cells , resulting in a high accuracy and lower latency . a cognitive radio network ( crn )
@xcite is a collection of cognitive radio nodes ( or processors ) , called secondary users ( sus ) that exploit the existing spectrum opportunistically .
the sus have the _ leira _ ( learning , efficiency , intelligence , reliability , and adaptively ) property for scanning and operating on multiple heterogeneous channels ( or frequency bands ) in the absence of the licensed user(s ) , termed as primary user(s ) ( pus ) , of the respective bands @xcite .
each pu has a fixed bandwidth , high transmit power , and high reliability ; however , the sus work on a broad range of bandwidth with low transmit power and low reliability .
a crn in 5 g networks is used for designing multi - tier architectures , removing interference among cells , and minimizing energy consumption in the network @xcite .
a crn creates a 2-tier architecture , similar to architectures discussed in section [ subsec : two - tier architectures ] ; however , it is assumed that either a mbs or a sbs has cognitive properties for working on different channels .
hong et al .
@xcite presented two types of crn - based architectures for 5 g networks , as : ( _ i _ ) non - cooperative and ( _ ii _ ) cooperative crns .
the _ non - cooperative crn _ establishes a multi - rats system , having two separate radio interfaces that operate at the licensed and temporary unoccupied channels by pus , called cognitive channels .
the sus work only on cognitive channels and form a crn , which overlays on the existing licensed cellular network .
the two networks can be integrated in the upper layers while must be separated in the physical layer .
this architecture can be used in different manners , as : ( _ i _ ) the cognitive and licensed channels are used by users near a mbs and users far away from the mbs , respectively , ( _ ii _ ) the cognitive and licensed channels are used for relaxed qos and strict qos , respectively .
the _ cooperative crn _ uses only a licensed channel , where sus access the channel in an opportunistic fashion when the pu of the channel is absent .
this architecture can be used in different manners , as : ( _ i _ ) a sbs communicates with a mbs using the licensed channel and provides service to its ues via an opportunistic access to the licensed frequency band , ( _ ii _ ) a licensed channel is used to serve ues by a sbs and the opportunistic access to the licensed channel is used to transfer backhaul data to the mbs . in short , the cooperative crn @xcite provides a real intuition of incorporating crns in 5 g networks , where a sbs works as a su , which scans activities of a macrocell and works on temporarily unoccupied frequency bands ( known as _ spectrum holes _
@xcite ) by a pu to provide services to their ues with minimally disrupting macrocell activities .
a dynamic pricing model based on a game theoretic framework for cognitive small - cells is suggested in @xcite . since in reality
sbss operators and mbss operators may not be identical and small - cells ues may achieve a higher data rate as compared to macrocells ues , the pricing model for both ues must be different .
huang et al .
@xcite provided an approach for avoiding inter - tier interference by integrating a _ cognitive technique _ at a sbs .
the cognitive technique consists of three components , as : ( _ i _ ) a cognitive module , which senses the environment and collects information about spectrum holes , collision probability , qos requirements , macrocell activities , and channel gains , ( _ ii _ ) a cognitive engine , which analyzes and stores the collected information for estimating available resources , and ( _ iii _ ) a self - configuration module , which uses the stored information for dynamically optimizing several parameters for efficient handoff , interference , and power management .
further , the channel allocation to a small - cell is done in a manner to avoid inter - tier and intra - tier interferences , based on gale - shapley @xcite spectrum sharing scheme , which avoids collisions by not assigning an identical channel to neighboring small - cells .
wang et al .
@xcite suggested an approach for mitigating inter - tier interference based on spectrum sensing , spectrum sharing , and cognitive relay , where links between a mbs and its ues are considered as pus and links between a sbs and its ues are considered as sus .
cognitive techniques are used for detecting interference from a mbs to a sbs and vice versa , and a path loss estimation algorithm is provided for detecting interference from a small - cell s ues to a macrocell s ues . after detecting inter - tier interference , a small - cell shares spectrum with a macrocell using either _ overlay spectrum sharing scheme _ ( _ i_._e_. , sus utilize unoccupied channels , and it is applicable when a mbs and a sbs s ues are very close or no interference is required by a macrocell s ues ) or _ underlay spectrum sharing scheme _
( _ i_._e_. , sus and pus transmit on an identical channel while restricting transmit power of sus , and hence , resulting in a higher spectrum utilization ) . note
that a crn can be used to support d2d communication and mitigate interferences caused by d2d communication , which we will see in section [ subsec : device - to - device communication architectures ] . * advantages of crns in 5 g networks .
* * _ minimizing interference _ : by implementing a crn at small - cells , cognitive small - cells can avoid interference very efficiently by not selecting identical channels as the channels of neighboring small - cells . * _ increase network capacity _ : the spectrum holes can be exploited for supporting a higher data transfer rate and enhancing bandwidth utilization @xcite .
* open issues . *
usually , cellular networks are not energy - efficient as they consume energy in circuits , cooling systems , and also radiate in air .
hence , an energy - efficient deployment of a crn in a cellular network is of utmost importance @xcite .
further , there is a tradeoff between the spatial frequency reuse and the outage probability , which requires the selection of an efficient spectrum sensing algorithm @xcite .
device - to - device ( d2d ) communication allows close proximity ues to communicate with each other on a licensed cellular bandwidth without involving a mbs or with a very controlled involvement of a mbs .
the standards and frameworks for d2d communication are in an early stage of research . in this section
, we will review d2d communication networks in short . for
a detailed review of d2d communication , interested readers may refer to @xcite . * challenges in d2d communication .
* * _ interference management _ : ues involved in d2d communication , say d - ues , face ( or create ) interference from ( or to ) other ues , or from ( or to ) a bs , based on the selection of a dl or ul channel , respectively .
the following types of interferences are investigated in @xcite : * * when using a dl channel : ( _ i _ ) interference from bss in the same cell , ( _ ii _ ) interference from other co - channel d - ues in the same cell , and ( _ iii _ ) interference from bss and co - channel d - ues from other cells . * * when using a ul channel : ( _ i _ ) interference from all co - channel c - ues in the same cell and other cells , and ( _ ii _ ) interference from all co - channel d - ues in the same cell and other cells .
+ _ proposed solutions _ : a simple solution may exist by implementing crns in d2d communication , as : d - ues are considered as sus and c - ues are considered as pus that should not be interfered .
consequently , any mechanism of crns can be implemented in d2d communication for interference removal . * _ resource allocation _ : when ues involved in d2d communication , it is required to allocate a sufficient amount of resources , particularly bandwidth and channels .
however , the allocation of optimum resources to d - ues must be carried out in a fashion that c - ues do not have interference from d - ues , and d - ues can also communicate and exchange data efficiently @xcite . + _ proposed solutions _
: sara @xcite , frame - by - frame and slot - by - slot channel allocation methods @xcite , and a social - aware channel allocation @xcite . *
_ delay - sensitive processing _ : audio , video streaming , and online gaming , which are natural in close proximity ues , require real - time and delay - sensitive processing .
hence , it is required to consider delay - sensitive and real - time processing in d2d communication @xcite .
+ _ proposed solutions _ : solutions based on channel state information ( csi ) and qos are provided in @xcite . *
_ pricing _ : sometimes a d - ue uses resources ( _ e_._g_. , battery and data storage ) of other ues for relaying information , where the other ue may charge for providing its resources .
hence , the design of a pricing model is needed , thereby a d - ue is not charged more money than that involved to communicate through a mbs . + _ proposed solutions _ : some solutions based on game theory , auction theory , and bargaining are suggested in @xcite .
* d2d communication types .
* d2d communication can be done in the following four ways @xcite , as follows : 1 . _ device relaying with operator controlled link establishment _
( dr - oc ) : a ue at the edge of a cell or in a poor coverage area can communicate with a mbs by relaying its information via other ues , which are within the stronger coverage area and not at the edge ; see figure [ fig : two - tier architecture for 5 g networks with small - cells ] .
2 . _ direct d2d communication with operator controlled link establishment _ ( dc - oc ) : source and destination ues
communicate directly with each other without involving a mbs , but they are _ assisted _ by the mbs for link establishment ; see figure [ fig : two - tier architecture for 5 g networks with small - cells ] .
device relaying with device controlled link establishment _ ( dr - dc ) : source and destination ues communicate _ through a relay _ without involving a mbs , and they are also responsible for link establishment ; see figure [ fig : two - tier architecture for 5 g networks with small - cells ] .
4 . _ direct d2d communication with device controlled link establishment _ ( dc - dc ) : source and destination ues communicate _ directly _ with each other without involving a mbs , and they are also responsible for link establishment ; see figure [ fig : two - tier architecture for 5 g networks with small - cells ] .
note that dr - oc and dc - oc involve a mbs for resource allocation and call setup , and hence , prevent interference among devices to some extent .
r10 cm [ fig : d2d communication architecture based on a social networking ] two types of coding schemes ( or communication types ) are described in @xcite : ( _ i _ ) two - way relay channel ( trc ) , where a source and a destination communicate through a relay , and ( _ ii _ ) multiple - access relay channel ( mrc ) , where multiple sources communicate to a destination through a relay with direct links .
note that the workings of dr - oc and mrc , and dr - dc and trc are identical .
two types of node discovery and d2d communication methods are also studied in @xcite , namely network - controlled approach and ad hoc network approach that work in a similar manner as dc - oc and dc - dc , respectively . * resource allocation methods .
* now , we will review an architecture and some methods for resource allocation in d2d communication . _
social - aware d2d architecture _ : as d2d communication is very efficient for close proximity ues , keeping this fact in view , li et al .
@xcite suggested a social - aware d2d communication architecture based on the social networking .
the architecture , see figure [ fig : d2d communication architecture based on a social networking ] , has four major components as : * _ ties _ : they are similar to friend relations in a social media , and hence , may be used as a trust measurement between two ues . allocating more spectrum and energy resources to ues with strong ties
can increase the peer discovery ratio , avoid congestion , and improve spectral efficiency . * _
community _ : it is similar to a group on facebook and helps in allocating more resources to all the ues in a community to decrease content duplication and increase the network throughput . * _ centrality _ : it is similar to a node that has more communication links / friends in a social network .
the concept of centrality in d2d communication reduces congestion and increases the network throughput by allocating more resources to a central node . *
_ bridges _ : they are similar to a connection between two communities .
hence , two devices forming a bridge can be allocated more resources as compared to other devices . _ channel allocation methods _ : two cooperative channel allocation methods , _ frame - by - frame _ and _ slot - by - slot _ , are given in @xcite .
consider three zones @xmath6 , @xmath7 , and @xmath8 with some ues such that @xmath6 and @xmath7 , @xmath7 and @xmath8 intersect , but @xmath6 and @xmath8 do not intersect , and @xmath7 holds a ue that communicates with ues of @xmath6 and @xmath8 . in the frame - by - frame channel allocation method , ues of @xmath6 and @xmath8 do intra - zone communication at different frames , and the ues of @xmath7 also communicate at different frames .
however , in the slot - by - slot channel allocation method , ues of @xmath6 and @xmath8 do intra - zone communication at an identical time , and of course , ues of @xmath7 communicate at a different time .
both the methods improve the efficiency of frequency division multiplexing and increase the network throughput .
hoang et al .
@xcite provided an iterative algorithm for subcarrier and power allocation such that the minimal individual link data rates and proportionate fairness among d2d links are obtained .
a 2-phase service - aware resource allocation scheme , called _ sara _ , is proposed @xcite . in the first phase of sara
, resources are allocated on - demand to meet different service requirements of d - ues , and in the second phase of sara , the remaining resources are allocated to d - ues such that the system throughput increases .
wang et al .
@xcite provided a delay - aware and dynamic power control mechanism that adapts the transmit power of d - ues based on instantaneous values of csi , and hence , finds the urgency of the data flow .
the dynamic power control selects a power control policy so that the long - term average delay and the long - term average power cost of all the flows minimize .
* advantages of d2d communication .
* d2d communication results in link reliability among d - ues , a higher data rate to d - ues , instant communication , an easy way for peer - to - peer file sharing , local voice services , local video streaming , local online gaming , an improved spectral efficiency , decreased power consumption of d - ues , and the traffic offload from a mbs .
* open issues .
* * _ security and privacy _ : in d2d communication , d - ues may take helps from other ues as relay nodes ; hence , it is required to communicate and transfer data in secure and privacy - preserving manners . consequently , the designing of energy - efficient and trust - making protocols is an open issue . * _ network coding scheme _ :
when d2d communication uses relay nodes , an efficient network coding scheme may be utilized for improving the throughput @xcite . *
_ multi - mode selection _ : in the current design of d2d communication , ues can do either d2d communication or communication to a bs @xcite ; however , it is not efficient . hence , there is a need to design a system that allows a ue to engage the two types of communication modes ( _ i_._e_. , d2d communication and communication to a bs ) simultaneously .
r7.5 cm [ fig : c - ran - whole - cloud ] cloud computing @xcite infrastructure provides on - demand , easy , and scalable access to a shared pool of configurable resources , without worrying about the management of resources .
the inclusion of the cloud in the mobile cellular communication can provide its benefits to the communication system . in this section
, we will see cloud - based architectures or cloud - based radio access networks ( c - rans ) for 5 g networks .
a detailed review of c - rans is given in @xcite .
* the main idea of a c - ran .
* the first c - ran is provided by china mobile research institute @xcite . the basic idea behind
any c - ran is to execute most of the functions of a mbs in the cloud , and hence , divide the functionality of a mbs into a _ control layer _ and a _ data layer _ ; see figure [ fig : c - ran - whole - cloud ] .
the functions of the control and the data layers are executed in a cloud and in a mbs , respectively .
thus , a c - ran provides a _ dynamic _ service allocation scheme for scaling the network without installing costly network devices . specifically , a mbs has two main components , as : ( _ i _ ) a baseband unit ( bbu , for implementing baseband processing using baseband processors ) , and ( _ ii _ ) a remote radio head ( rrh , for performing radio functions ) . in most of the c - rans ,
bbus are placed in the cloud and rrhs stay in mbss .
thus , a c - ran provides an easily scalable and flexible architecture .
we will see advantages of c - rans at the end of this section . * challenges in the deployment of a c - ran .
* * _ an efficient fronthaul data transfer technique _ : a flexible cloudification of the functions of a mbs comes at the cost of efficient fronthaul data transfer from rrhs to bbus .
the fast and efficient data transfer to the cloud has a proportionate impact on the performance of a c - ran @xcite . *
_ real - time performance _ : since c - rans will be used instead of a mbs that provides all the services to users , it is required to transfer and process all the data in the cloud as fast as a mbs can do ; otherwise , it is hard to find solutions to real - time problems using a c - ran @xcite . *
_ reliability _ : the cloud provider does not ensure any guarantee of failure - free executions of their hardware and software .
thus , it is hard to simulate an error - free mbs using a c - ran . *
_ security _ : the resources of the cloud are shared among several users and never be under the control of a single authority .
hence , a malicious user may easily access the control layer of a c - ran , resulting in a more severe problem . *
_ manageability _ : it is clear that a non - secure c - ran may be accessed by any cloud user , which poses an additional challenge in manageability of c - rans .
further , the dynamic allocation of the cloud resources at a specific time interval is a critical issue ; otherwise , a c - ran may face additional latency @xcite .
now , we will see some c - ran architectures in brief .
* 2-layered c - ran architectures . * the authors @xcite provided two c - ran architectures based on the division of functionalities of a mbs , as : ( _ i _ ) full centralized c - ran , where a bbu and all the other higher level functionalities of a mbs are located in the cloud while a rrh is only located in the mbs , and ( _ ii _ ) partially centralized c - ran , where a rrh and some of the functionalities of a bbu are located in the mbs while all the remaining functions of the bbu and higher level functionalities of the mbs are located in the cloud . thus , the authors @xcite proposed the use of only two layers , namely a control layer and a data layer for implementing c - rans , as follows : 1 .
_ data layer _ : it contains heterogeneous physical resources ( _ e_._g_. , radio interface equipment ) and performs signal processing tasks ( _ e_._g_. , channel decoding , demultiplexing , and fast fourier transformation ) .
2 . _ control layer _ : it performs baseband processing and resource management ( application delivery , qos , real - time communication , seamless mobility , security , network management , regulation , and power control ) ; see figure [ fig : c - ran - whole - cloud ] .
rost et al .
@xcite introduced ran - as - a - service ( ranaas ) concept , having the control and the data layers .
however , in ranaas , a cloud provides flexible and on - demand ran functionalities ( such as network management , congestion control , radio resource management , medium access control , and physical layer management ) , according to the network requirements and characteristics , unlike @xcite .
hence , there is no need to split functionalities in advance to the control and the data layers , as a result ranaas provides more elasticity . till now , it is clear that how a c - ran will work .
however , in order to achieve real - time performance , a rrh executing latency - critical applications may connect to a nearby small cloud while other rrhs that are not adhered to real - time applications may connect to a far larger cloud @xcite .
softair @xcite is also a two - layered c - ran that performs mobility - management , resource - efficient network virtualization , and distributed and collaborative traffic balancing in the cloud .
* 3-layered c - ran architectures . *
the full - centralized c - ran architecture @xcite has some disadvantages , as : continuous exchange of raw baseband samples between the data and the control layers , and the control layer is usually far away from the data layer resulting in a processing delay . in order to remove these disadvantages , liu et al .
@xcite proposed _ convergence of cloud and cellular systems _ ( concert ) . in this architecture , one more layer ,
called a _ software - defined service layer _ , is introduced at the top of the control layer .
the functioning of the layers in concert is as follows : 1 .
_ data layer _ : is identical to the full centralized c - ran s data layer @xcite , having rrhs with less powerful _ computational resources _ for application level computations .
control layer _ : works just as a logically centralized entity .
the control layer coordinates with the data layer resources and presents them as virtual resources to the software - defined service layer .
the control layer provides a few services as : radio interfacing management , wired networking management , and location - aware computing management to the data layer .
software - defined services layer _ : works as a virtual bs and provides services ( _ e_._g_. , application delivery , qos , real - time communication , seamless mobility , security , network management , regulation , and power control ) to the data layer .
wu et al .
@xcite enhanced c - ran architecture @xcite and ranaas @xcite , by moving the whole ran to a cloud .
the proposed architecture also has three layers , where the data layer and the control layer are same to the respective layers of c - ran @xcite . the third layer , called a _ service layer _ , executes in the cloud and provides some more functionalities than the software - defined services layer of @xcite , _
e_._g_. , traffic management , the cell configuration , interference control , allocation of functional components to the physical elements , and video streaming services .
the authors @xcite proposed an all - software - defined network using three types of hierarchical network controllers , namely mbs controller , ran controller , and network controller , where except the mbs controller all the others can be executed in the cloud , as follows : 1 .
_ mbs controller _ : usually stays nearby ues , and performs wireless resource management and packet creation .
2 . _ ran controller _ : stays at the top of mbs controllers , and performs connectivity , rat selection , handoff , qos , policies , mobility management .
network controller _ : stays at the top of ran controllers , ensures end - to - end qos , and establishes application - aware routes .
* advantages of c - rans in 5 g networks .
* c - rans provide a variety of services as a software , power efficient , flexible , and scalable architecture for the future cellular communication . here , we enlist some advantages of c - rans , as follows : * _ an easy network management _ : c - rans facilitate on - demand installation of virtual resources and execute cloud - based resources that dynamically manage interference , traffic , load balance , mobility , and do coordinated signal processing @xcite . * _ reduce cost _ : it is very costly and time - consuming to deploy and install a mbs to increase the network capacity .
however , the deployment of c - rans involves less cost , while it provides usual services like a mbs @xcite . as a result
, operators are required to only deploy , install , and operate rrhs in mbss . * _ save energy of ues and a mbs _ : c - rans offload data - intensive computations from a mbs and may store data of ues and mbss .
consequently , c - rans allow ues and mbss to offload their energy - consuming tasks to a nearby cloud , which saves energy of ues and mbss @xcite . *
_ improved spectrum utilization _ : a c - ran enables sharing of csi , traffic data , and control information of mobile services among participating mbss , and hence , results in increased cooperation among mbss and reduced interference @xcite .
* open issues . * transferring data from rrhs to bbus , _
i_._e_. , from the data layer to the control layer , is a crucial step based on the selection of the functions of a mbs that has to be sent to a cloud , resulting in the minimal data movement in the network @xcite .
however , the selection of functions to be executed in a cloud and a mbs is a non - trivial affair @xcite .
the security and privacy issues involved in the cloud computing effect c - rans , and hence , the development of a c - ran has to deal with inherent challenges associated with the cloud and the wireless cellular communication simultaneously .
energy - efficient infrastructures are a vital goal of 5 g networks .
researchers have proposed a few ways of reducing energy in the infrastructure .
rowell et al .
@xcite considered a joint optimization of energy - efficiency and spectral - efficiency .
a user - centric 5 g network is suggested in @xcite so that ues are allowed to select ul and dl channels from different bss depending on the load , channel conditions , services and application requirements . in a similar manner , decoupling of signaling and
data is useful for energy saving ; for example , a mbs may become a signaling bs while sbss may serve all data requests .
thus , when there is no data traffic in a sbs , it can be turned off . a similar approach for decoupling of signaling and data is presented in @xcite .
however , in @xcite , a ue gets connected to a sbs according to instructions by a mbs , and hence , it results in less energy consumption at ues side due to less interference , faster small - cells discovery , and mbs - assisted handover .
hu and qian @xcite provided an energy - efficient c - ran architecture in a manner that rrhs serve almost a same number of ues .
they also present an interference management approach so that the power consumption of sbss and mbss can be decreased . like rowell et al .
@xcite , hu and qian @xcite also suggested that the association of a ue can not be done based on entirely a dl channel or a ul channel , and a ue must consider both the channels at the time of association with a bs .
lin et al .
@xcite suggested to include an energy harvesting device ( to collect energy ) and a spectrum harvesting controller ( to collect spectrum ) at sbss .
in this section , we will see issues regarding the interference , handoff , qos , load balancing , and channel access management in the context of 5 g networks .
we have already seen challenges in interference management ( section [ section : challenges in the development of 5 g ] ) . in this section
, we will review some techniques / methods for interference management in 5 g networks .
nam et al .
@xcite handled ue - side interference by using a new type of receiver equipment , called an advanced receiver , which detects , decodes , and removes interference from receiving signals .
in addition , the network - side interference is managed by a joint scheduling , which selects each ue according to the resources needed ( _ e_._g_. , time , frequency , transmission rate , and schemes of multiple cells ) for its association with a bs .
hence , the joint scheduling , which can be implemented in a centralized or distributed manner , requires a coordination mechanism among the neighboring cells .
hossain et al .
@xcite proposed distributed cell access and power control ( capc ) schemes for handling interference in multi - tier architectures .
capc involves : ( _ i _ ) prioritized power control ( ppc ) , which assumes that ues working under a sbs have a low - priority than ues working under a mbs , and hence , low - priority ues set their power so that the resulting interference must not exceed a predefined threshold ; ( _ ii _ ) cell association ( ca ) , which regards dynamic values of resources , traffic , distance to a mbs , and available channels at a mbs for selecting a mbs with the optimum values of the parameters ; and ( _ iii _ ) resource - aware ca and ppc ( rca - ppc ) , which is a combination of the first two approaches and allows a ue to connect simultaneously with multiple bss for a ul channel and a dl channel based on criteria of ppc and ca .
hong et al .
@xcite suggested to use self - interference cancellation ( sic ) in small - cells networks .
as we have seen that sbss require methods to transfer backhaul data to a mbs ( section [ subsubsec : backhaul data transfer from small - cells ] ) , the use of sic can eliminate the need of such methods and result in _ self - backhauled small - cells _ ( sscs ) .
sscs use sic for providing services and backhaul data transfer , and more importantly , they gain almost the same performance as having a small - cell connected with a wired optical fiber .
it works as : in the dl channel , a sbs may receive from a mbs and simultaneously transmit to ues . in the ul channel ,
a sbs may receive from ues and simultaneously transmit data to the mbs .
therefore , a small - cell can completely remove the need of a separate backhaul data transfer method , resulting in reduced cost .
the authors @xcite suggested that the measurement of inter - user interference channel , and then , the allocation of ul and dl channels by a mbs can mitigate _
inter - user ul - to - dl interference _ in a _ single - cell _ full duplex radio network .
however , in the case of a _ multi - cell _ full duplex radio network , interference mitigation becomes more complex , because of the existence of interference in ul and dl channels between multi - cells ues that work on identical frequency and time .
* open issues .
* interference cancellation in a full duplex radio network is still an open problem for multi - cell .
the design of algorithms for inter - bs dl - to - ul - interference and inter - user ul - to - dl - interference cancellation in a full duplex radio network is still open and to be explored .
handoff provides a way to ues connected to a bs to move to another bs without disconnecting their sessions .
* challenges in the handoff process in 5 g networks . *
the handoff management in 5 g networks has inherent challenges associated with the current cellular networks , _
e_._g_. , minimum latency , improved routing , security , and less uncertainty of having no services .
the network densification , very high mobility , the zero latency , and accessing multi - rats make handoff management in 5 g networks harder .
also , the current cellular networks do not provide efficient load balancing for a bs at the time of handoff .
for example , movement of ues from houses to offices in the morning creates a load imbalance at bss of respective areas .
* types of handoff in 5 g networks . *
three types of handoffs are presented in the context of 5 g networks , as follows : 1 .
_ intra - macrocell handoff _ : refers to handoff between small - cells that are working under a single mbs @xcite .
inter - macrocell handoff _ : refers to handoff between macrocells .
it may also lead to handoff between two small - cells that are working under different mbss .
note that if the handover between small - cells of two different mbss is not done properly , then the inter - macrocell handoff also fails @xcite .
multi - rats handoff _ : refers to handoff of a ue from a rat to other rat .
song et al .
@xcite provided a handoff mechanism for highly mobile users , where a ue sends some parameters ( _ e_._g_. qos , signal - to - interference ratio ( sir ) , and time to handoff ) in a measurement report to the current mbs .
sir is considered as a primary factor for finding a situation for an initiation of the handoff .
the gray system model predicts the @xmath9 measurement report from the @xmath10 measurement report .
the predicted value is used for the final decision for the handover process .
zhang et al .
@xcite proposed a handoff mechanism assisted by a mbs .
the mbs collects several parameters from ues , and if the mbs finds the values of the received parameters below a threshold , then it finds a new sbs or mbs for handoff and informs to the ues . for handoff over different rats , orsino et al .
@xcite proposed a handoff procedure so that a ue can select the most suitable rat without any performance loss .
a ue collects received signal strength ( _ i_._e_. , rsrp ) or quality ( _ i_._e_. , rsrq ) from the current mbs , and then , it initiates handoff if rsrq is below a threshold .
the ue collects several parameters ( _ e_._g_. , transmitted power , the cell s traffic load , and ue requested spectral efficiency ) from adjacent bss , and then , selects the most suitable bs .
duan et al .
@xcite provided an authenticated handoff procedure for c - rans and multi - tier 5 g networks .
the control layer holds an authentication handover module ( ahm ) for monitoring and predicting the future location of ues ( based on the current location ) and preparing relevant cells before ues arrival in that .
the ahm holds a master public - private key pair with each rrh , which are authenticated by the ahm in off - peak hours , and ues are verified before accessing the network services by rrhs .
ues sends i d , the physical layer s attributes , location , speed , and direction to the control layer in a secure manner for the handoff process .
the proposed approach reduces the risk of impersonation and man - in - the - middle attacks .
giust et al .
@xcite provided three distributed mobility management protocols , the first is based on the existing proxy mobile ipv6 ( pmipv6 ) , the second is based on sdn , and the third is based on routing protocols .
* open issues .
* handoff mechanisms are yet to be explored
. it will be interesting to find solutions to extremely dense hetnets .
furthermore , the handoff process may also create interference to other ues ; hence , it is required to develop algorithms while considering different types of interferences in 5 g networks and a tradeoff between the number of handoffs and the level of interference in the network @xcite .
we have already seen challenges in qos management in section [ section : challenges in the development of 5 g ] . in this section
, we will review some techniques / methods for qos management in 5 g networks .
zhang et al .
@xcite provided a mechanism for different delay - bounded qos for various types of services , applications , and users having diverse requirements , called heterogeneous statistical delay - bounded qos provisioning ( hsp ) .
hsp maximizes the aggregate effective capacity of different types of 5 g architectures , reviewed in sections [ subsec : two - tier architectures ] , [ subsec : cognitive radio network based architectures ] , and [ subsec : device - to - device communication architectures ] .
hsp @xcite algorithm claims better performance over other approaches ; however , it imposes new challenges in terms of the assignment of different resources for different links under the cover of delay - bounded qos requirements .
the authors @xcite proposed the deployment of a quality management element ( qme ) in the cloud for monitoring inter - ues and inter - layer ( the control and the data layers in c - rans ) qos .
rrus send wireless information ( _ e_._g_. , csi , reference signal received quality , and resource block utilization ) to the qme .
consequently , the qme executes service control algorithms ( to manage qos and some other activities like traffic offloading and customized scheduling ) , and then , sends scheduling strategies to the rrus to achieve a desired level of qos .
kim et al .
@xcite provided a routing algorithm for multi - hop d2d communication .
the algorithm takes into account different qos for each link so that it can achieve better performance than max - min routing algorithms .
the algorithm increases flow until it provides the desired qos or reaches the maximum capacity of the link .
since the algorithm considers individual links , there is a high probability that some of the channels will serve multiple - links with the desired qos .
zhou et al .
@xcite provided a qos - aware and energy - efficient resource allocation algorithm for dl channels , where ues are allocated an identical power in one case and non - identical power in the second case .
the algorithm maximizes energy - efficiency while minimizes transmit power .
hu and qian @xcite suggested that a ue must consider the source data rate , delay bound , and delay - violation probability before connecting to a mbs . * open issues . *
the 5 g networks are supposed to satisfy the highest level of qos .
the tactile internet @xcite requires the best qos , especially , latency of the order of 1 millisecond for senses such as touching , seeing , and hearing objects far away , as precise as human perception .
however , the current proposed architectures do not support efficient tactile internet services . in future
, it would be a promising area as to encode senses , exchange data satisfying the zero latency , and enable the user to receive the sensation .
load balancing means allocation of resources to a cell such that all the users meet their demands .
it is an important issue in the cellular wireless networks . in a 2-tier architecture , discussed in section [ subsec : two - tier architectures ] , user offloading to a small - cell is useless if there is no resource partitioning @xcite .
singh and andrews @xcite provided an analytical and tractable framework for modeling and analyzing joint resource partitioning and offloading in a 2-tier architecture .
hossain et al .
@xcite provided a technique for cell association based on dynamic resources and traffic in a cell , as we have already discussed in section [ subsec : interference management in 5 g networks ] . in a fast moving vehicle , _
e_._g_. , a train , it is very hard to allocate resources without any service interruption at all .
a distributed load balancing algorithm for fast moving vehicles is presented in @xcite .
load balancing architectures for d2d communication are given in @xcite , which we discussed in section [ subsec : device - to - device communication architectures ] .
interested readers may refer to @xcite for further details of load balancing in 5 g networks .
channel access protocols allow several ues to share a transmission channel without any collision while utilizing the maximum channel capacity . * challenges in channel access control management in 5 g networks .
* channel access control management in 5 g networks faces inherent challenges associated with the current cellular networks , _
e_._g_. , synchronization , fairness , adaptive rate control , resource reservation , real - time traffic support , scalability , throughput , and delay .
in addition , providing the currently available best channel in 5 g networks is prone to additional challenges , as : high mobility of ues , working at the higher frequencies ( @xmath5 3 ghz ) , different rats , dense networks , high qos , high link reliability , and the zero latency for applications and services .
the authors @xcite proposed a frame - based medium access control ( fd - mac ) protocol for mmwave - based small - cells .
fd - mac consists of two phases , as : ( _ i _ ) scheduling phase , when a sbs collects the traffic demands from the supported ues and computes a schedule for data transmission , and ( _ ii _ ) transmission phase , when the ues start concurrent transmissions following the schedule .
a schedule , which is computed using a graph - edge coloring algorithm , consists of a sequence of topologies and a sequence of time intervals for indicating how long each topology should sustain .
liu et al .
@xcite provided mac protocols for ues of small - cells and macrocells .
two types of mac protocols for sbs s ues are suggested , as : contention - based random channel access ( crca ) , where ues randomly access the channel and send messages if the channel is available , and reservation - based channel access ( rca ) , which uses time division multiple access . for macrocells ues , they provided a mac protocol for crn - based 5 g networks , where sus sense a licensed channel until it is free or the residual energy of sus exceeds a predetermined threshold , for saving their battery .
further , they evaluated a tradeoff between the network throughput and sensing overhead .
the authors @xcite suggested two crn - based channel access techniques for cognitive sbss .
the first channel access scheme is termed as contention - resolution - based channel access ( cca ) , which is similar to crca @xcite , is based on carrier sense multiple access .
the other channel access scheme is termed as uncoordinated aggressive channel access ( uca ) , which works identically as rca @xcite , for aggressively using channels in a small - cell for increasing opportunistic spectrum access performance .
nikopour et al .
@xcite provided a multi - user sparse code multiple access ( mu - scma ) for increasing dl s spectral efficiency .
mu - scma does not require the complete csi , and hence , provides high data rate and the robustness to mobility .
nikopour et al .
@xcite also provided an uplink contention based scma for massive connectivity , data transmission with low signaling overhead , low delay , and diverse traffic connectivity requirements .
* open issues .
* the current channel access protocols do not regard qos and latency challenges in 5 g networks .
hence , there is a scope of designing algorithms for finding multiple reliable links with the desired qos and the zero latency . in this section ,
we present security and privacy related challenges and a discussion of security and privacy protocols in the context of 5 g networks .
* challenges in security and privacy in 5 g networks . *
_ authentication _ is a vital issue in any network . due to the zero latency guarantee of 5 g networks , authentication of ues and network devices
is very challenging , since the current authentication mechanisms use an authentication server that takes hundreds of milliseconds delay in a preliminary authentication phase @xcite .
a _ fast and frequent handover _ of ues over small - cells requires for a robust , efficient , and secure handoff process for transferring context information @xcite .
security to _ multi - rats _
selection is also challenging , since each rat has its own challenges and certain methods to provide security ; clearly , there is a need to provide overlapped security solutions across the various types of rats . _
c - rans _ also inherit all the challenges associated with the cloud computing and wireless networks .
in addition , several other challenges ( _ e_._g_. , authorization and access control of ues , availability of the network , confidentiality of communication and data transfer , integrity of communication and data transmission , accounting and auditing of a task , low computation complexity , and communication cost ) require sophisticated solutions to make a secure 5 g network . the security and latency are correlated as a higher level of security and privacy results in increased latencies .
therefore , the communication satisfying the zero latency is cumbersome when combined with secure and privacy - preserving 5 g networks .
monitoring is suggested for securing the network and detecting intruders @xcite . however , monitoring a large number of ues ( by a trusted authority ) is not a trivial task ; hence , we do not see monitoring as a preferred way to secure networks .
yang et al .
@xcite focused on the physical layer security , which is independent of computational complexity and easily handles any number of devices .
the physical layer security protocol considers locations of ues and provides the best way for ues for securely selecting a mbs or a sbs without overloading the network .
tehrani et al .
@xcite provided a method for secure and private d2d communications , called close access , where d - ues have a list of other trusted d - ues devices , and all such ues can communicate directly using an encryption scheme while the remaining ues not in the list utilize a mbs - assisted communication .
an encryption based video sharing scheme is also presented in @xcite .
kantola et al .
@xcite proposed a policy - based scheme that can prevent dos and spoofing attacks . a solution to secure handoff
is given in @xcite and discussed in section [ subsec : handoff management in 5 g networks ] .
* open issues .
* the current security and privacy solutions to 5 g networks are not impressive and , hopefully , unable to handle massive connections .
we can clearly visualize a potential scope for developing _ latency - aware protocols _ along with security awareness that must consider secure data transmission , end - to - end security , secure and private storage , threats resistant ues , and valid network and software access .
it is already mentioned in section [ section : introduction ] that the development of 5 g networks requires the design and implementation of new methodologies , techniques , and architectures .
we have reviewed some of the methodologies and technologies in the previous sections , such as : ( _ i _ ) full duplex radios in sections [ section : challenges in the development of 5 g ] and [ subsec : interference management in 5 g networks ] , ( _ ii _ ) crns in section [ subsec : cognitive radio network based architectures ] , ( _ iii _ ) d2d communication in section [ subsec : device - to - device communication architectures ] , ( _ iv _ ) multi - tier heterogeneous deployment or dense - deployment techniques in sections [ subsec : two - tier architectures ] , [ subsec : cognitive radio network based architectures ] , and [ subsec : device - to - device communication architectures ] , ( _ v _ ) c - rans in section [ subsec : cloud - based architectures ] , ( _ vi _ ) ` green ' communication systems in section [ subsec : energy - efficient architectures for 5 g networks ] , and ( _ vii _ ) techniques related to interference , qos , handoff , channel access , load balancing in section [ sec : management issues in 5 g networks ] . in this section , we will briefly describe some techniques that are mentioned but not explained in earlier sections .
* self - interference cancellation ( sic ) .
* when a full duplex radio receives signals from another radio , it also receives interference signals by its own transmission , resulting in self - interference .
hence , a full duplex radio has to implement techniques to cancel self - interference @xcite .
sic techniques are classified into passive and active cancellations ; see @xcite . as advantages ,
the implementation of sic enables seamless global roaming , high - throughput services , and low - latency applications in a cost effective manner @xcite .
* downlink and uplink decoupling ( dud ) .
* in the current cellular networks , a ue is associated with a bs based on the received signal power in its dl channel , and then , uses the same bs for ul channel transmission @xcite .
dud allows a ue to select a dl channel and a ul channel from two different bss , based on the link quality , the cell load , and the cell backhaul capacity @xcite .
therefore , a ue may have the dl channel connected through a bs and the ul channel connected through a different bs , resulting in a user - centric 5 g architecture @xcite and improving the capacity of ul channels , which is a prime concern .
* network function virtualization ( nfv ) . *
nfv @xcite implements network functions such as network address translation , firewalls , intrusion detection , domain name service , the traffic load management , and caching through software running on commodity servers .
however , the conventional networks implement these functions on dedicated and application specific servers .
hence , nfv decreases the burden on network operators by not updating dedicated servers / hardware , thereby saves cost .
* software - defined networking ( sdn ) . * sdn architectures @xcite partition network control functions and data forwarding functions , thereby the network control functions are programmable , and the network infrastructure handles applications and network services .
sdn architectures can be divided into three parts , as : ( _ i _ ) _ the software controller _ : holds network control functions such as the network manager , apis , network operating system , and maintaining the global view of the network ; ( _ ii _ ) _ the southbound part _ : provides an interface and a protocol between the controller and sdn - enable infrastructure , where openflow @xcite is the most famous protocol that provides communication between the controller and the southbound part ; ( _ iii _ ) _ the northbound part _ : provides an interface between sdn applications and the controller @xcite .
interested readers may refer to @xcite to find details of sdn , challenges in sdn , and applications of sdn .
note that sdn , nfv , and c - rans offload functionalities to software running on commodity servers .
however , sdn separates network control functions from data forwarding functions , while nfv implements network functions in software @xcite . besides that c - rans integrate both sdn and nfv to meet the scalability and flexibility requirements in the future mobile networks @xcite . * millimeter waves ( mmwave ) . *
the current wireless bandwidth is not able to support a huge number of ues in 5 g networks .
hence , researchers are looking at 30 ghz to 300 ghz frequency bands , where mmwave communication is proposed for achieving high - speed data transfer .
the current research focuses on 28 ghz band , 38 ghz band , 60 ghz band , and the e - band ( 7176 ghz and 8186 ghz ) @xcite .
however , mmwave has several challenges at the physical , mac , and network layers .
there are a number of papers about mmwave , and hence , we are not discussing mmwave in details .
interested readers may refer to @xcite . * machine - to - machine ( m2 m ) communication . *
m2 m communication refers to the communication between ( network ) devices ( _ e_._g_. , sensors / monitoring - devices with a cloud ) without human intervention .
some examples of m2 m communication are intelligent transport systems , health measurement , monitoring of buildings , gas and oil pipelines , intelligent retail systems , and security and safety systems .
however , the development of m2 m communication involves several challenges to be handled in the future , as follows : connectivity of massive devices , bursty data , the zero latency , scalability in terms of supporting devices , technologies and diverse applications , fast and reliable delivery of messages , and cost of devices .
in addition , the development of efficient algorithms for location , time , group , priority , and multi - hop data transmission management for m2 m communication is needed @xcite .
interested readers may refer to @xcite .
* massive mimo ( mmimo ) .
* mmimo systems are also known as large - scale antenna systems , very large mimo , hyper - mimo , and full - dimension mimo @xcite .
mmimo systems use antenna arrays with hundreds of antennas at mbs for simultaneously serving many ues with a single antenna in identical time and frequency .
hence , expensive equipment are mounted on a mbs @xcite .
a mmimo system relies on spatial multiplexing , which in turn relies on the channel knowledge at mbs , on both ul and dl channels .
a mmimo system reduces latency and energy , simplifies the mac layer , shows robustness against intentional jamming , and increases the capacity due to spatial multiplexing . * visual light communication ( vlc ) .
* vlc is a high - speed data transfer medium for short range los optical links in the future cellular networks @xcite .
the light - emitting diodes ( leds ) provide vlc using amplitude modulation at higher frequencies and achieve higher data rates while keeping the led s primary illumination function unaffected .
vlc can be used for outdoor applications , where high power laser - based equipment provide transmission links , and for indoor application , where leds provide short distance transmission links .
vlc is an energy - efficient technology , works on a wider range of unregulated frequency bands , shows high spatial reuse , and inherits security due to los .
however , vlc is sensitive to sunlight and not able to work for a long range without los , and hence , confined coverage @xcite .
the implementation of vlc has still some unanswered questions , such as : how vlc will work in a long range without los and will it work for backhaul data transfer in multihop ? * fast caching .
* caching is a way for storing temporary data for reducing data access from slow memory or the network . in a network ,
content caching is popular and answers the request while responding in place of the application servers as a proxy , thereby reducing the amount of hits that are directly sent to the ultimate backend server .
in caching , three decisions are prominent as : what to cache , where to cache , and how to cache ?
the authors @xcite suggested that ues will have enough memory in the future , and they can work as a cache for any other ue , since a small amount of popular data requires to be cached .
wang et al .
@xcite provided a caching mechanism based content centric networking ( ccn ) , assuming 5 g networks will include ccn - capable gateways , routers , and mbss .
ccn provides in - network data storage , also known as _
universal caching _ , at every node in the network .
the cached data is uniquely identified at each node .
accordingly , a user can request for a particular content from the content cache of any device within the network , or the request is forwarded to the actual source of content storage .
interested readers may refer to @xcite for finding details about some of the above mentioned methodologies and technologies .
the zero latency , high speed data transfer , and ubiquitous connectivity are the salient features of 5 g networks that are expected to serve a wide range of applications and services . in this section
, we enumerate the most prominent applications of 5 g networks , as follows : * personal usages . *
this domain of 5 g networks would be capable of supporting a wide range of ues , from scalable to heterogeneous devices . also the data demands ( _ e_._g_. , multimedia data , voice communication , and web surfing ) , would be satisfied while keeping the qos requirements
. * virtualized homes . * due to c - ran architectures , users may have only low cost ues ( _ e_._g_. , set - top box for tvs and residential gateways for accessing the internet ) with services of the physical and data link layers .
all the other higher layers applications may move to the cloud for universal access and outsourced computation services @xcite .
* smart societies .
* it is an abstract term for connected virtualized homes , offices , and stores .
accordingly , every digital and electronic services / appliances , _
e_._g_. , temperature maintenance , warning alarms , printers , lcds , air conditioners , physical workout equipment , and door locks , would be interconnected in a way that the collaborative actions would enhance the user experience .
similarly , smart stores would assist in filtering out irrelevant product details , sale advertisements , and item suggestions on the go .
* smart grids .
* the smart grids would decentralize the energy distribution and better analyze the energy consumption .
this would allow the smart grids to improve efficiency and economic benefits .
the 5 g networks would allow a rapid and frequent statistical data observation , analysis , and fetching from remote sensors and would adjust the energy distribution accordingly . * the tactile internet . *
the tactile internet improves the user experience in a virtual environment to an extent of only milliseconds of interaction latency .
the futuristic applications such as automated vehicle platooning , self - organizing transportation , the ability to acquire a virtual sense for physically challenged patients , synchronized remote smart - grids , remote robotics , and image processing with customized / panoramic view would use the tactile internet protocols @xcite . * automation . *
self - driving vehicles would take place in the near future , and as a requirement , vehicles would communicate with each other in real - time .
moreover , they would communicate with other devices on the roads , homes , and offices with a requirement of almost zero latency .
hence , an interconnected vehicular environment would provide a safe and efficient integration with other information systems . * healthcare systems .
* a reliable , secure , and fast mobile communication can strengthen medical services , _
e_._g_. , frequent data transfer from patients body to the cloud or health care centers .
therefore , the relevant and urgent medical services could be predicted and delivered to the patients very fast .
* logistics and tracking . * the future mobile communication would also assist in inventory or package tracking using location based information systems .
the most popular way would be to embed a radio frequency identification ( rfid ) tag and to provide a continuous connectivity irrespective of the geographic locations .
* industrial usages . *
the zero latency property of 5 g networks would help robots , sensors , drones , mobile devices , users , and data collector devices to have real - time data without any delay , which would help to manage and operate industrial functions quickly while preserving energy .
in this section , we present some real demonstrations and testbeds for 5 g networks .
docomo @xcite is developing a real - time simulator for evaluating and simulating mmimo , small - cell , and mmwave .
the simulator has shown 1000-times increase in the system capacity , and 90% users achieved 1gbps data rate .
docomo also performed a real experiment in the year 2012 , where data was uploaded at the speed of 10gbps .
samsung performed data transmission experiment using mmwave at 28 ghz frequency band and achieved the world s first highest data rate of 1.2gbps on a vehicle running at the speed of @xmath5 100 km/h .
further , when the vehicle was nearby a stop , the data transmission speed was achieved up to 7.5gbps . in the experiment ,
the peak data rate was more than 30-times faster as compared to the state - of - the - art 4 g technology .
samsung is also developing array antennas that have nearly zero - footprint and reconfigurable antenna modes @xcite .
ericsson has achieved the speed of 5gbps in a demonstration .
huawei at university of surrey in guildford is developing a testbed , which would be used for developing methodologies , validating them , and verifying a c - ran for an ultra - dense network .
the european commission funded metis @xcite has developed more than 140 technical components , including : air interface technologies , new waveforms , multiple access and mac schemes , multi - antenna and mmimo technologies , multi - hop communications , interference management , resource allocation schemes , mobility management , robustness enhancements , context aware approaches , d2d communication , dynamic reconfiguration enablers , and spectrum management technologies .
they have also implemented some testbeds , as : three d2d communication related testbeds , one massive machine communications related testbed , and one related to waveform design .
there are other projects funded by the european commission working on the development of 5 g networks , as : 5gnow ( http://www.5gnow.eu/ ) , tropic ( www.ict-tropic.eu ) , mcn ( www.mobile-cloud-networking.eu ) , combo ( www.ict-combo.eu ) , moto ( www.fp7-moto.eu ) , and phylaws ( www.phylaws-ict.org ) . _ small - cells _ :
sprint , verizon , and at&t in the united states , vodafone in europe , and softbank in japan are developing femtocells @xcite .
alcatel - lucent , huawei , and nokia siemens networks has been involved in the development of plug - and - play sbss @xcite .
_ mmimo _ : titanmimo by nutaq is a testbed for 5 g mmimo .
the titanmimo-4 testbed provides a realistic throughput by aggregating the entire mmimo channel into a central baseband processing engine .
lund university and national instruments , austin are also involved in the development of testbeds for mmimo @xcite .
docomo @xcite is developing a real - time simulator for mmimo .
_ c - rans _ : china mobile research institute has developed c - rans @xcite .
the european commission supported ijoin ( http://www.ict-ijoin.eu/ ) is also involved in developing c - ran architectures , especially , ranaas @xcite .
ibm , intel , huawei , and zte are developing c - rans @xcite .
openairinterface ( oai ) @xcite is an open experimentation and prototyping platform created by the mobile communications department at eurecom .
oai used for two purposes as : c - rans and m2 m communication .
_ full duplex radio _ :
groups at stanford and rice universities are focusing on the development of full duplex radios @xcite .
_ crns _ : vodafone chair mobile communications systems at the tu dresden has created a testbed for studying crns in the future cellular networks @xcite . _
mmwave _ : docomo @xcite , huawei , the european commission funded projects , and new york university @xcite are developing methodologies and testbeds for mmwave and have successfully performed various experiments .
in this survey , we discussed salient features , requirements , applications , and challenges involved in the development of the fifth generation ( 5 g ) of cellular mobile communication that is expected to provide very high speed data transfer and ubiquitous connectivity among various types of devices .
we reviewed some architectures for 5 g networks based on the inclusion of small - cells , cognitive radio networks , device - to - device communication , and cloud - based radio access networks .
we find out that energy consumption by the infrastructure is going to be a major concern in 5 g networks , and hence , reviewed energy - efficient architectures .
we figured out several open issues , which may drive the future inventions and research , in all the architectures .
the development of new architectures is not only a concern in 5 g networks ; there will be a need for handling other implementation issues in the context of users , _
e_._g_. , interference removal , handoff management , qos guarantee , channel accessing , and in the context of infrastructures , _
e_._g_. , load balancing . during our illustration
, we included several new techniques , _
e_._g_. , full duplex radios , dense - deployment techniques , sic , dud , mmwave , mmimo , and vlc .
we also discussed the current trends in research industries and academia in the context of 5 g networks based on real - testbeds and experiments for 5 g networks .
we conclude our discussion with a resonating notion that the design of 5 g infrastructure is still under progress .
the most prominent issues are enlisted below , and to provide elegant solutions to these issues would contribute in early deployment as well as in long run growth of 5 g networks . * the security and privacy of devices , infrastructures , communication , and data transfer is yet to be explored .
we believe that the current solutions based on encryption would not suffice in the future due to a huge number of devices
. intuitively , a solution that would use an authenticated certificate may be feasible @xcite .
* the development of network devices , infrastructures , and algorithms must be self - healing , self - configuring , and self - optimizing to preform dynamic operations as per the need , for example , dynamic load balancing , qos guarantee , traffic management , and pooling of residual resources . *
the cloud computing is an attractive technology in the current trend for various applications .
we have reviewed c - rans ; however , the current solutions do not consider the impact of virtualization for backhaul data transfer , the trust of the cloud , inter - cloud communication , ubiquitous service guarantee , and real - time performance guarantee with zero - latency .
thus , the development of c - rans must address the big question that how much virtualization is good ? * the multi - rats are attractive solutions to access different rats
. however , would it be possible for devices to use more than one rat at an identical time for uplink and downlink channels ?
further , the network densification must quantify that how much network density is good ? * the design , development , and usage of user devices , service - application models , and , especially , the network devices must be affordable to cater the needs of overwhelming users , service providers , and network providers .
* the zero latency is a primary concern in most of the real - time applications and services , especially , in the tactile internet .
however , all the existing architectures and implementations of 5 g networks are far from achieving the zero latency .
therefore , it is utmost desired that the latest real - time and ultra - reliable network configurations must be improved to a latency free environment . _ more than 50 billion connected devices
_ , ericsson , white paper , 2011 .
available at : http://www.akos-rs.si/files/telekomunikacije/digitalna_agenda/internetni_protokol_ipv6/more-than-50-billion-connected-devices.pdf .
_ the next generation of communication networks and services _ , the 5 g infrastructure public private partnership ( 5gppp ) , european commission , 2015 .
available at : http://5g-ppp.eu/wp-content/uploads/2015/02/5g-vision-brochure-v1.pdf .
available at : http://research.microsoft.com/en-us/projects/spectrum/fcc_dynamic_spectrum_access_noi_comments.pdf .
[ http://research.microsoft.com/en-us/projects/spectrum/fcc_dynamic_spectrum_access_noi_comments.pdf .
] _ requirements and vision for ng - wireless _ , outlook : visions and research directions for the wireless world , wireless world research forum , 2011 .
available at : http://www.wwrf.ch/files/wwrf/content/files/publications/outlook/outlook7.pdf .
n. bhushan , j. li , d. malladi , r. gilmore , d. brenner , a. damnjanovic , r. sukhavasi , c. patel , and s. geirhofer .
network densification : the dominant theme for wireless evolution into 5 g .
, 52(2):8289 , 2014 .
h. elshaer , f. boccardi , m. dohler , and r. irmer .
load & backhaul aware decoupled downlink / uplink access in 5 g systems . in _ 2015 ieee international conference on communications , icc 2015 , london , united kingdom , june 8 - 12 , 2015 _ , pages 53805385 , 2015 . c. x. mavromoustakis , a. bourdena , g. mastorakis , e. pallis , and g. kormentzas . an energy - aware scheme for efficient spectrum utilization in a 5 g mobile cognitive radio network architecture .
, 59(1):6375 , 2015 .
a. orsino , g. araniti , a. molinaro , and a. iera .
effective rat selection approach for 5 g dense wireless networks . in _
ieee 81st vehicular technology conference , vtc spring 2015 , glasgow , united kingdom , 11 - 14 may , 2015 _ , pages 15 , 2015 .
a. osseiran , f. boccardi , v. braun , k. kusume , p. marsch , m. maternia , o. queseth , m. schellmann , h. d. schotten , t. hidekazu , h. m. tullberg , m. a. uusitalo , b. timus , and m. fallgren .
scenarios for 5 g mobile and wireless communications : the vision of the metis project . , 52(5):2635 , 2014 .
t. s. rappaport , s. sun , r. mayzus , h. zhao , y. azar , k. wang , g. n. wong , j. k. schulz , m. samimi , and f. g. jr .
millimeter wave mobile communications for 5 g cellular : it will work ! , 1:335349 , 2013 .
j. vieira , s. malkowsky , k. nieman , z. miers , n. kundargi , l. liu , i. wong , v. wall , o. edfors , and f. tufvesson .
a flexible 100-antenna testbed for massive mimo . in _
ieee globecom workshops _ , pages 287293 , 2014 . c. wang , f. haider , x. gao , x. you , y. yang , d. yuan , h. m. aggoune , h. haas , s. fletcher , and e. hepsaydir .
cellular architecture and key technologies for 5 g wireless communication networks . , 52(2):122130 , 2014 .
y. zhou , d. li , h. wang , a. yang , and s. guo .
-aware energy - efficient optimization for massive mimo systems in 5 g . in _
sixth international conference on wireless communications and signal processing , wcsp _
, pages 15 , 2014 . | [ abs : abstrat ] the rapidly increasing number of mobile devices , voluminous data , and higher data rate are pushing to rethink the current generation of the cellular mobile communication . the next or fifth generation ( 5 g ) cellular networks
are expected to meet high - end requirements .
the 5 g networks are broadly characterized by three unique features : ubiquitous connectivity , extremely low latency , and very high - speed data transfer .
the 5 g networks would provide novel architectures and technologies beyond state - of - the - art architectures and technologies . in this paper ,
our intent is to find an answer to the question : `` _ _ what will be done by 5 g and how _ _ ? ''
we investigate and discuss serious limitations of the fourth generation ( 4 g ) cellular networks and corresponding new features of 5 g networks .
we identify challenges in 5 g networks , new technologies for 5 g networks , and present a comparative study of the proposed architectures that can be categorized on the basis of energy - efficiency , network hierarchy , and network types .
interestingly , the implementation issues , _ e_._g_. , interference , qos , handoff , security - privacy , channel access , and load balancing , hugely effect the realization of 5 g networks .
furthermore , our illustrations highlight the feasibility of these models through an evaluation of existing real - experiments and testbeds .
: cloud radio access networks ; cognitive radio networks ; d2d communication ; dense deployment ; multi - tier heterogeneous network ; privacy ; security ; tactile internet . |
glitches , observed from more than 30 isolated pulsars @xcite , are a sudden jump in pulsar rotation rate . slow recovery of a rotation rate following these events implies the presence of loosely - coupled fluids or neutron superfluids in the interior of a neutron star .
the transfer of angular momentum from the rotating neutron superfluid to the normal matter is likely to be responsible for pulsar glitches and postglitch relaxation .
a rotating neutron superfluid is threaded by quantized vortex lines .
the superfluid can alter its angular velocity by the radial motion of vortices . in the inner crust of a neutron star neutron rich nuclei that form the bcc lattice
coexist with neutron superfluid .
the vortex lines in the inner crust are normally pinned to the lattice nuclei .
difference in velocity between superfluid and nuclear lattice builds up since magnetic braking slows down the nuclear lattice and pinning prevents the vortex lines from moving .
then , the neutron star undergoes sudden unpinning of a large number of vortex lines followed by the outward motion .
this catastrophic unpinning has long been considered as a promising cause for glitches @xcite .
a vortex line is subjected to the magnus force when it moves relative to the superfluid .
the catastrophic unpinning model assumes that the pinning force is strong enough to sustain the vortices to the nuclear lattice against the magnus force until the moment just before glitches .
based on the condensational and kinetic energies , the pinning energy is estimated , ranging approximately @xmath2 mev per nucleus @xcite . this magnitude itself is strong enough against the magnus force expected just before glitches .
jones ( 1992 ) mentions , however , that the pinning forces on a randomly oriented rigid vortex line largely cancel since there are nearly an equal number of pinning sites on either side of the vortex line .
link , epstein , & baym ( 1993 ) find that for a vortex line of finite tension , pinning becomes much more efficient by slightly bending and forming kinks .
recently , jones ( 1997 , 1998 , 1999 ) argues that vortex interaction with a polycrystalline structure does not provide pinning strong enough to explain the large glitches observed in the vela pulsar ( see section [ pinning ] for detail ) .
here we study the vortex configurations , oscillations and pinning in the inner crust of a neutron star . in section [ equation ]
we derive the equation of motion of vortex lines . in section [ configuration ]
we present the equilibrium configurations of vortex lines . in section [ oscillation ]
we examine the stability of equilibrium configurations and the oscillations excited on a vortex line . in section [ pinning ] we discuss the vortex pinning based on the results of the previous sections . in the last section we summarize the results and mention the conclusions .
we consider the motion of vortex lines in an inner crust , where the lattice nuclei are embedded in a superfluid neutron sea .
we adopt the cartesian coordinates fixed in the rest frame of the lattice nuclei .
the z - axis is chosen to be parallel to one of the major axes of a body - centered crystal lattice . in this paper
the most attention will be devoted to configurations and oscillations of vortex lines which are parallel to or slightly inclined against the z - axis .
we limit the inclination angle of a vortex line against the z - axis to a small value .
then , we can describe their motion and configurations by the two - dimensional position vectors , @xmath3 , of vortex lines in the xy plane which depend on the coordinate z. the equation of motion of vortex lines is written as @xmath4 where @xmath5 is the superfluid density , @xmath6 the pinning force per unit length of a vortex line , @xmath7 the circulation of a vortex , @xmath8 the locally uniform superfluid velocity and @xmath9 the tension of the vortex line .
vortices in the superfluid are quantized .
each vortex line carries one quantum @xmath10 of circulation , where @xmath11 is the planck constant and @xmath12 the neutron mass . since we consider the vortex lines aligned or almost aligned with the z - axis , we approximate the circulation vector by @xmath13 where @xmath14 is the unit vector along the z - axis . in equation ( [ eq : eqmot ] ) the second term represents the magnus force which arises when there exists the relative motion between the local superfluid and the vortex .
the magnus force acting on a unit length of a vortex line is given by @xmath15 where @xmath16 is the vortex velocity .
in the vortex unpinning model the difference in angular velocity between the inner crust superfluid and the crust is reduced significantly just after the glitches , decoupling the inner crust superfluid from the crust .
as the crust is slowed down by the external braking torque , the angular velocity difference @xmath17 builds up between the local superfluid and the lattice nuclei . using the glitch interval @xmath18 and the angular deceleration rate @xmath19 , the angular velocity difference
just before the glitch is estimated as @xmath20 in the vela pulsar the angular velocity difference and the magnus force acting on the pinned vortex line may amount to @xmath21 and @xmath22 , respectively , just before the glitches .
the third term in equation ( [ eq : eqmot ] ) comes out from the tension of vortices when vortices are distorted from the straight lines .
the energy of vortex lines is equal to the kinetic energy of the velocity field induced by them .
the tension is given by the energy per unit length , @xmath23 where the parameter @xmath24 is only logarithmically dependent on the wave number and the neutron superfluid coherence length @xcite .
for numerical estimates we assume @xmath25 @xcite .
the vortex - nucleus interaction has an attractive component arising from the change in the condensation energy and a repulsive component arising from the change in kinetic energy of circulating superfluid .
we adopt only the attractive component since we are interested in the inner portion of the inner crust , where the attractive component is dominant over the repulsive component .
the interaction potential we adopt is @xmath26 where @xmath27 is the pinning energy ( or the interaction energy ) per nucleus , @xmath28 is the vortex core radius or the coherence length of the neutron superfluid and @xmath29 is the distance in the xy plane between the nucleus and the vortex line , @xmath30 . the coherence length of the neutron superfluid is written as @xmath31 where @xmath32 is the energy gap for the superfluid neutrons , @xmath33 their fermi energy and @xmath34 the fermi surface wave number @xcite . the pinning force , which acts to pin the vortex line to nuclei , per site is approximately given by @xmath35
we can obtain the lattice - vortex interaction potential by summing the above potential ( [ eq : enpi ] ) contributed from each nucleus over the entire nuclei of a single large crystal .
note that the lattice - vortex interaction potential has the translational properties and hence can be expanded into the fourier series using the reciprocal lattice vectors .
we assume the lattice - vortex interaction potential to be uniform in the z - direction since we are interested in the configurations and oscillations of vortices , which are approximately aligned with the z - axis , with the scale and wave lengths significantly larger than the lattice constant .
hence , of the twelve basic reciprocal lattice vectors of the bcc lattice we adopt the basic four vectors in the xy plane , @xmath36 and @xmath37 , where @xmath38 with @xmath39 being the lattice constant .
then , the lattice - vortex interaction potential per unit length of a vortex line is given by @xmath40,\ ] ] where the fourier coefficient @xmath41 is @xmath42 at densities of @xmath43 .
we obtain the pinning force acting on the unit length of a vortex line by taking the derivative of @xmath44 with respect to @xmath45 and @xmath46 , @xmath47 \label{eq : fx}\ ] ] and @xmath48,\label{eq : fy}\ ] ] where @xmath49 and @xmath50 are the @xmath45 and @xmath46 components of the pinning force , respectively .
we conduct the coordinate rotation and transform the variables from @xmath51 to @xmath52 defined as @xmath53 and @xmath54 using equations ( [ eq : fx ] ) , ( [ eq : fy ] ) , ( [ eq : phi ] ) and ( [ eq : psi ] ) , the equation of motion of vortices ( [ eq : eqmot ] ) reduces to @xmath55 and @xmath56 here and hereafter we assume for simplicity that the neutron superfluid flows in the y - direction . note that equations ( [ eq : eqmot1 ] ) and ( [ eq : eqmot2 ] ) describe the motion of vortices in terms of the components in the directions of reciprocal vectors .
if the vortex lines are completely rigid , the vortex configuration would be simply a straight line .
the tension of a superfluid vortex line in the inner crust of a neutron star , however , is finite .
hence , when the vortex lines undergo the forces such as the pinning and magnus forces , the vortex lines in general are bent and their shape is deformed from the straight line .
we examine the equilibrium configurations of vortex lines here .
we assume that in equilibrium @xmath57 and @xmath58 in equations ( [ eq : eqmot1 ] ) and ( [ eq : eqmot2 ] ) and hence the vortices are stationary with respect to the nuclear lattice .
then , equations ( [ eq : eqmot1 ] ) and ( [ eq : eqmot2 ] ) can be integrated to yield @xmath59 and @xmath60 once a displacement and orientation of the vortex line are given at , for example , z = 0 , we can solve equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) and derive the equilibrium configurations of vortex lines .
hereafter , for numerical calculations we use the physical parameters such as the pinning energy , the coherence length and the lattice constant at densities of @xmath61 , where a large fraction of crust superfluids resides and the pinning energy peaks .
we adopt @xmath62 for the superfluid velocity unless otherwise stated .
the first , second and third terms in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) correspond to the pinning force , the magnus force and the vortex tension , respectively .
the pinning force per site is @xmath63 dyne and the vortex tension is @xmath64 dyne , while the magnus force acting on the vortex section of a length corresponding to the lattice constant @xmath39 is @xmath65 dyne .
we note that the magnus term is much smaller than the pinning and tension terms .
we should note that equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) are analogous to the energy conservation equation for the particle motion in a potential well .
the first and second terms and third term on the left hand side of equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) may correspond to the potential and kinetic energies in the particle motion , respectively , while the integration constant on the right hand side to the total energy .
the gradient of the vortex line against the z - axis may be looked as the velocity of a particle .
the difference between the `` total and potential energies '' yields the gradient of the vortex line .
the pinning term in equation ( [ eq : stst - int1 ] ) ( equation ( [ eq : stst - int2 ] ) ) has peaks at @xmath66 ( @xmath67 ) with @xmath68 being integer , while the magnus term increases ( decreases ) linearly with @xmath69 ( @xmath70 ) . as easily inferred from this analogy , solutions to equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) can be classified into the three characteristic cases in terms of the magnitude of the integration constant ( the `` total energy '' ) relative to the peak values of the potential energy due to the pinning and magnus forces .
let us consider the case where the integration constant ( the `` total energy '' ) in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) is just equal to a peak value of the sum of pinning and magnus terms ( the `` potential energy '' ) .
the derivative of @xmath69 and @xmath70 with respect to z should be zero over an entire vortex line .
hence , the equilibrium solution for the vortex configuration is a straight line which is in parallel with the z - axis or the major axis of a single crystal lattice ( fig .
[ straight ] ) .
note that the average orientation of a vortex line expresses the direction of the rotation axis .
thus , the rotation axis of the star is also parallel to the z - axis .
when the magnus term is neglected in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) , the straight vortex line passes through the bottom of the pinning potential well along the z - axis . when the magnus term is included , the straight vortex line , though still parallel to the z - axis , is displaced only slightly from the z - axis , @xmath71 and @xmath72 . in actual cases when the magnus force acts on the vortex lines , however ,
the equilibrium configuration of the vortex may deviate from the straight line , though very slightly , since the pinning sites are located discretely along the z - axis .
we have the strongest pinning regime when the rotation axis happens to be oriented along the major axis of a single crystal since the vortex line intercepts the pinning sites most often .
let us move to the case where the integration constant ( the `` total energy '' ) in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) is smaller than peak values of the sum of pinning and magnus terms ( the `` potential energy '' ) .
this case corresponds to the motion of a particle which is bound in the potential well .
as the particle exhibits a periodic motion , the vortex line goes back and forth between the two neighboring lattice planes ( fig .
[ sinepmt ] ) .
the average direction of the vortex line and hence the rotation axis of the star are parallel to the z - axis .
note that the significant energy is required to deform the vortex from the straight line to the sinusoid - like shape ( see the next sub - section for the calculation of the total energy of a vortex ) .
hence , it may be the straight vortex line which is realized in nature when the rotation axis is parallel to the z - axis or the major axis of a single crystal lattice .
for simplicity and clarity we neglect the magnus force here .
let us consider the case where the integration constant ( the `` total energy '' ) in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) is larger than the peak value of the pinning term ( the `` potential energy '' ) .
this case corresponds to the motion of a particle which continues to move in one direction with varying speed .
the derivative of @xmath69 and @xmath70 with respect to z is not now zero everywhere along a vortex line .
hence , we examine here the equilibrium vortex configuration when the average orientation of a vortex line and hence the rotation axis of the star are inclined against the z - axis ( or the major axis of a single crystal ) .
we study first the equilibrium solutions when the angle between the rotation axis of a star and the major axis of a crystal is infinitesimally small . over an entire interval of a large single crystal the vortex passes from one lattice plane to the adjacent lattice plane .
hence , we set the boundary conditions such as @xmath73 at @xmath74 and @xmath75 at @xmath76 .
then , we can solve equations ( [ eq : stst - int1 ] ) analytically and derive the equilibrium configurations , @xmath77 \label{eq : kink},\ ] ] where @xmath78 figure [ kink ] shows that the vortex can move from one lattice plane to the adjacent lattice plane by bending itself and forming a kink when the vortex tension is finite .
the vortex line is displaced substantially from the pinning sites in the kink part , while in the other part the vortex line is straight and passes through nearly all of the pinning sites along the z - axis .
@xmath79 given by equation ( [ eq : kinkl ] ) expresses a length of the kink . in the middle of a kink the vortex line
is inclined against the z - axis by @xmath80 .
energy is needed to create a kink from a straight vortex line since the length of a vortex line increases and a part of it should overcome the pinning potential .
the self - energy of vortex lines is equal to the kinetic energy of the velocity field induced by them .
the increase of the vortex self - energy due to bending is @xmath81 dz,\end{aligned}\ ] ] while the energy necessary to overcome the pinning potential is @xmath82 the length of a kink can be determined from minimizing the sum of @xmath83 and @xmath84 @xcite . using the solution ( [ eq : kink ] ) together with @xmath85 ,
the total energy necessary to form a kink is calculated as @xmath86 which yields @xmath87 at densities of @xmath88 .
we increase gradually the integration constant in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) and hence the inclination angle @xmath89 of the rotation axis against the major axis ( z - axis ) of a single crystal .
the vortex lines now should thread through many lattice planes .
we find that when the inclination angle is small and less than @xmath90 , vortex lines in equilibrium configurations move from one lattice plane to the next by forming a clear kink and the number of kinks increases with increasing inclination angle ( fig .
[ stpt ] ) .
the separation @xmath91 in z between the neighboring kinks varies as @xmath92 .
the property of a kink in fig .
[ stpt ] is the same as that in fig .
a vortex line can orient its average direction to the direction of the rotation axis by adjusting a number of kinks .
apart from the kink parts , the vortex lines are straight and in parallel to the z - axis , and are firmly pinned to the lattice nuclei .
even though the extra energy is required to create kinks , the total energy of a vortex line with kinks is much lower than that of the inclined straight vortex line without kinks since except for kink parts the vortex line lies in the bottom region of the pinning potential @xcite . as the average orientation of a vortex line inclines against the major axis , the separation between the neighboring kinks and hence the portion of a vortex line available for pinning decrease . as seen in fig .
[ stpt ] , the straight pinned part of the vortex line becomes shorter than the kink part when the inclination angle is above @xmath93 .
the vertical pinned portion as well as the kink feature becomes less prominent at @xmath94 .
when the inclination angle exceeds @xmath95 , the equilibrium configuration of a vortex line becomes almost the straight line .
this consequence arises from the fact that the portion of a vortex line for pinning becomes too short to compensate the energy cost for producing kinks when the inclination angle exceeds a certain value .
when the average orientation of a vortex line inclines significantly against the z - axis , we need to include the z - components for both the force and displacement , though neglected here .
we expect that the lattice planes different from those considered here may become important for pinning .
the three dimensional calculations are needed in order to know the equilibrium configuration of a vortex line accurately .
we now examine how the kink solutions are modified when we include the magnus terms in equations ( [ eq : stst - int1 ] ) and ( [ eq : stst - int2 ] ) .
the pinning and magnus terms as a function of @xmath69 ( @xmath70 ) are represented by the cosine curve and the inclined straight line , respectively .
hence , there exists @xmath96 ( @xmath97 ) at which the integration constant ( the `` total energy '' ) becomes equal to the sum of the pinning and magnus terms ( the `` potential energy '' ) .
the integration constant is larger ( smaller ) than the sum of the pinning and magnus terms at @xmath98 ( @xmath99 ) , while smaller ( larger ) at @xmath100 ( @xmath101 ) .
this case corresponds to the motion of a particle which moves forwards and then at the potential wall bounces to move backwards .
figure [ kink - parab ] illustrates the equilibrium configurations of vortex lines for different superfluid velocities @xmath102 relative to the lattice nuclei .
the equilibrium configuration is characterized locally by a kink and globally by a parabola . as explained above
, the kink arises from the balance between the pinning and tension terms .
as expected from the analogy with the particle motion , the vortex lines extends first rightwards and then leftwards as z increases . when we take a long scale average of equation ( [ eq : eqmot ] ) , the pinning forces , which alternate the sign ( the direction ) along a vortex line ,
cancel out whereas the magnus force , which is constant along a vortex line , is left .
hence , when combined with the tension term in equation ( [ eq : eqmot ] ) , the constant magnus term yields the parabolic configuration .
the curvature radius of a vortex line is given approximately by @xmath103 as the superfluid velocity relative to the lattice nuclei increases , the curvature and hence the deformation due to the magnus force becomes larger ( fig .
[ kink - parab ] ) .
notice also that the kink feature becomes less prominent as the deformation becomes larger due to the same reason as mentioned in the previous sub - section .
the three dimensional calculations are needed again .
numerical estimates in this paper are made mostly by adopting the physical parameters at the inner part of an inner crust , where the density is higher and the pinning energy peaks .
the kink properties and pinning strength depend on matter densities through the nature of nuclei and superfluids .
for example , the kink length and creation energy are @xmath104 and @xmath105 at densities of @xmath43 in the inner part of an inner crust , while @xmath106 and @xmath107 at densities of @xmath108 in the outer part of an inner crust .
the choice of parameters may affect the results .
note , however , that most of crust superfluids reside in the inner part of an inner crust .
hence , even if the outer part of the inner crust is included adequately , the change may be small and may not alter the essence of our conclusions .
the inner crust nuclei are likely to take a polycrystalline structure @xcite .
the orientation of a crystal lattice may change from one domain to another . then , along a vortex line there may lie a large number of single crystals with various orientations against the average direction of a vortex line .
the important physical parameters which vary with the orientation of a crystal lattice are the distance between the neighboring lattice planes for pinning and the distance between the neighboring pinning sites along a vortex line .
it is likely that even when the vortex line is not close to the main axis of a crystal lattice , the equilibrium configuration of a vortex line may also be expressed by a kink solution with the pinning lattice planes different from those containing the main axis of a crystal lattice . as seen from equations ( [ eq : kinkl ] ) and
( [ eq : et ] ) , the kink length and creation energy are in proportion to the distance between the neighboring lattice planes for pinning and depend weakly on the fourier coefficient ( @xmath41 ) of a pinning potential .
the kink properties may vary , though not significantly , depending on the orientation of a crystal lattice .
the pinning strength of a vortex line is inversely proportional to the separation of the neighboring pinning sites along a vortex line .
when the vortex line is not close to the major axis of a crystal lattice , the pinning site distance will be longer and hence the pinning of a vortex line will be less strong .
the various orientations of crystal lattice along a vortex line may reduce the pinning strength of a vortex line as a whole compared to the case with a small inclination angle .
we expect , however , that pinning , though weaken , is still strong enough to explain the large glitches since the change of the pinning site distance more than an order of magnitude is less likely .
we need to conduct the three dimensional calculations in order to understand the equilibrium configuration and pinning strength accurately for various orientations of crystal lattice as well as for large magnus forces .
let us now consider the oscillations excited on and propagate along the vortex line @xcite . in order to examine oscillations of small amplitude about equilibrium
, we write @xmath109 and @xmath110 where the subscripts @xmath111 and @xmath112 denote the equilibrium solutions and the perturbations , respectively . using equations ( [ eq : phi1 ] ) and ( [ eq : psi1 ] ) , we linearize equations ( [ eq : eqmot1 ] ) and ( [ eq : eqmot2 ] ) and obtain @xmath113 and @xmath114 we consider plane - wave solutions @xmath115 for @xmath116 and @xmath117 , where @xmath118 is the angular frequency and @xmath119 is the wave number .
then , we derive the dispersion relation for oscillations , @xmath120 if we adopt such solutions as @xmath121 for equilibrium , from equation ( [ eq : disp ] ) we obtain the dispersion relation simplified as @xmath122 next , we examine the property of oscillations represented by the dispersion relation ( [ eq : disp2 ] ) . the first term in the dispersion relation ( [ eq : disp2 ] ) can be derived when in equations ( [ eq : pert1 ] ) and ( [ eq : pert2 ] ) we include only the second and third terms ( the magnus and tension terms ) ignoring the first term ( the pinning term ) .
hence , the first term in the dispersion relation ( [ eq : disp2 ] ) expresses the kelvin wave , which is circularly polarized and propagates helically along a vortex line @xcite . the second term in the dispersion relation ( [ eq : disp2 ] ) can be derived when in equations ( [ eq : pert1 ] ) and ( [ eq : pert2 ] ) we include only the first and second terms ( the pinning and magnus terms ) neglecting the third term ( the tension term ) .
we find that the second term in the dispersion relation expresses the circular motion of a vortex line as a whole around the equilibrium position .
in general the dispersion relation ( [ eq : disp2 ] ) represents the combined oscillations of kelvin and rotational modes .
figure [ disp ] illustrates the dispersion relation given by equation ( [ eq : disp2 ] ) for different @xmath123 .
the angular frequency of allowed oscillations is plotted as a function of wave number . at smaller wave numbers the angular frequency of oscillations
is determined mainly by the rotational mode , while at larger wave numbers mainly by the kelvin mode .
we should note that there exists the minimum frequency for the oscillations especially when @xmath123 is close to @xmath124 .
the minimum frequency for @xmath125 is @xmath126 which yields @xmath127 at densities of @xmath88 and @xmath128 at densities of @xmath129 .
@xmath130 in the pinned segments , where the vortex line lies deep in the pinning potential well . on the other hand , at the kink segments where @xmath131 , the minimum frequency is zero .
these results lead to the important conclusion that the vortex oscillations with frequencies lower than the minimum frequency , @xmath132 , can not be excited on and can not propagate in the pinned part of a vortex line , whereas oscillations with frequencies higher than the minimum can propagate along a vortex line , varying the wave length .
the vortex oscillations with lower frequencies are allowed in the kink part of a vortex line .
we should note , however , that the dispersion relation ( [ eq : disp2 ] ) in the kink part is less reliable for lower frequencies since the wave length exceeds the kink length and hence the local perturbation analysis is not adequate .
we need to conduct the non - local analyses for the accurate dispersion relation in low frequencies . in a polycrystalline structure
the vortex line passes through a number of single crystals with various orientations .
the local perturbation analysis can be applied to each single crystal to examine the vortex oscillations .
equations ( [ eq : disp2 ] ) and ( [ eq : wmin ] ) may also be derived for each single crystal , although the physical parameters in one domain differ from those in another .
we can expect the physical parameters to change rather sharply at the boundaries between the single crystals .
if the frequencies are sufficiently high above the minimum frequencies , the oscillations will propagate along a vortex line passing through many single crystals .
if the frequencies are around the minimum frequencies , the oscillations are reflected at the boundaries and in some cases will be trapped within one crystal or some .
if the frequencies are lower than the minimum frequencies , there may be no oscillations which propagate over a number of single crystals .
it has been considered that pulsar glitches are a consequence of the large scale unpinning of superfluid neutron vortices from the lattice nuclei in the inner crust of neutron stars .
recently , however , jones ( 1997 , 1998 , 1999 ) argues that vortex interaction with a polycrystalline structure does not provide pinning strong enough to explain the large glitches observed in the vela pulsar .
let us examine this conclusion in the light of our new results mentioned in the previous sections .
jones describes the reason for the conclusion that the structural regularity of a defect - free single crystal allows displacement of a vortex , under a magnus force , to a continuous sequence of new positions with no more than extremely small changes in energy .
furthermore , jones considers the motion of kinks propagating along a vortex line with a velocity of @xmath133 . for simplicity
we consider the vortex lines which have kinks bent in the y - direction and extend approximately in the z - direction .
the kink propagation along a vortex line yields a displacement and hence motion of the vortex line in the y - direction . if the kink propagation velocity is @xmath134 the time average of a vortex velocity in the y - direction is equal to the superfluid velocity @xcite . in effect , vortex lines move in the y - direction together with the superfluid , almost not suffering from pinning .
these energy and velocity arguments lead jones to the above conclusion that pinning would be extremely weak and insufficient for the large glitches .
we point out several difficulties included in these arguments .
kinks should be created and supplied around the edge of a single crystal at a rate of one kink every @xmath135 seconds in order for a vortex line to move steadily together with superfluid .
the kink supply rate required just before the glitches in the vela pulsar is @xmath136 .
this rate is much lower than the minimum frequency ( equation ( [ eq : wmin ] ) ) of oscillations which can be excited in the pinned part of a vortex line .
hence , the oscillations of vortex lines can not be a direct cause for the creation and supply of kinks .
moreover , energy of @xmath87 ( equation ( [ eq : et ] ) ) , which is substantially larger than the thermal energy , is needed locally to form a kink .
energy source for kink formation is also pointed out as a difficulty included in the kink propagation model .
we note that the kink solution ( [ eq : kink ] ) can also be expressed as a sum of fourier components of different wave numbers . as seen from the dispersion relation ( [ eq : disp2 ] ) , the phase velocity of a vortex wave depends on wave number .
the vortex waves with different wave numbers propagate with different phase velocities .
hence , even if a kink is formed around the end of a single crystal , the kink feature will be smeared out during propagation .
furthermore , our perturbation analyses show that the vortex configurations with static kink structures are stable .
these facts indicate that the kink motion following equation ( [ eq : vk ] ) is less likely .
if the kinks do not move along a vortex line with the velocity given by equation ( [ eq : vk ] ) , the conclusion by jones is not guaranteed .
we show in the previous sections that the vortex equilibrium configuration is in general composed of the static kink and straight segments .
a vortex line in equilibrium lies deep in the pinning potential well and is strongly pinned to the lattice nuclei in its most part except for the kink part . from the perturbation analyses we find no unstable mode that grows with time for the pinned equilibrium configurations both with and without kinks .
the pinned equilibrium configurations , though oscillations can be excited , may be stable . hence , the static kink is a stable structure .
our studies show that strong pinning of vortex lines is likely to occur especially when the average direction of a vortex line is parallel or approximately parallel to the major axis of a crystal lattice and the magnus force is not so strong .
the orientation of a crystal lattice varies from one domain to another in a polycrystalline structure .
the pinning strength decreases in the domain where the major axis of a crystal lattice is not close to the vortex line .
however , as long as the equilibrium vortex configurations are represented by the kink solutions , pinning of a vortex line in a polycrystalline structure would be still strong enough to explain the large glitches ( see the subsection 3.7 ) .
we have shown that the equilibrium configuration of a vortex line consists of straight and kink parts when the rotation axis of the star is approximately aligned with the major axis of a crystal lattice and the magnus force is weak .
the straight part is much longer than the kink part .
furthermore , the straight parts pass through nearly all of nuclei along the major axis , whereas the kink parts lie mostly in space between the lattice planes .
the vortex line is strongly pinned to the lattice nuclei in the straight part , while not in the kink part .
we have found from the perturbation analyses that there exist two modes , rotational and helical ( kelvin ) , for oscillations excited on a vortex line .
the vortex oscillations are possible only above a minimum frequency @xcite and the phase velocity of oscillations varies with wave number .
we find no unstable mode that grows with time .
these facts lead to the conclusion that the vortex configuration with kinks may be stable and yield the strongly pinned state especially when the vortex line is close to the main axis of a crystal lattice .
jones ( 1997 , 1998 ) argues that pinning is not strong enough to explain the large glitches adopting a kink motion in a polycrystalline structure .
our studies show that the static kink is a stable structure and the kink motion is less likely . if the kinks do not move along a vortex line , the conclusion made by jones is not guaranteed .
nuclei in the crust may be in a polycrystalline structure .
the orientation of a crystal lattice varies from one domain to another .
the pinning strength decreases in the domain where the major axis of a crystal lattice is not close to the vortex line . however ,
as long as the equilibrium vortex configurations are represented by the kink solutions , pinning of a vortex line in a polycrystalline structure would be still strong enough to explain the large glitches .
we have shown that the magnus force makes the vortex line parabolic on global scale and causes it to incline against the major axis of a crystal lattice .
as the magnus force increases and the rotation axis inclines against the major axis of a crystal lattice , we need to include the major axis components for force and displacement , though neglected in the present study .
we expect that the lattice planes different from those considered here may become important for pinning .
the pinning sites are discrete although we have assumed the uniform pinning potential along the major axis of a crystal lattice .
link , epstein and baym ( 1993 ) find that there appears the band structure in the energy spectrum for excitations on a vortex line when the discrete nature is included in the pinning site distribution .
it is also necessary to take account of the discrete nature of the pinning sites when we consider the vortex equilibrium configurations with the scale length comparable to the lattice constant .
the above arguments suggest that the three dimensional calculations adopting the discrete pinning sites should be conducted in near future in order to understand the equilibrium vortex configurations and vortex oscillations and judge the relevance of vortex pinning for the pulsar glitches correctly .
we thank m. ruderman , r. epstein and b. link for many useful discussions .
we also thank s. nakamura for his help in calculations .
this research was supported in part by the grant - in aid for scientific research ( c ) ( 10640234 , 12640229 , 12640302 ) .
alpar , m. a. 1977 , , 213 , 527 alpar , m. a. , anderson , p. w. , pines , d. , & shaham , j. 1984 , , 276 , 325 anderson , p. w. , & itoh , n. 1975 , , 256 , 25 baym , g. , epstein , r. i. , & link , b. 1992 , physica b , 178 , 1 epstein , r. i. , & baym , g. 1988 , , 328 , 680 epstein , r. i. , link , b , & baym , g. 1992 , in the structure and evolution of neutron stars , ed .
d. pines , r. tamagaki , & s. tsuruta ( reading : addison wesley ) jones , p. b. 1992 , , 257 , 501 jones , p. b. 1997 , , 79 , 792 jones , p. b. 1998 , , 296 , 217 jones , p. b. 1999 , , 306 , 327 link , b. , & epstein , r. i. , 1991 , , 373 , 592 link , b. , epstein , r. i. , & baym , g. 1993 , , 403 , 285 lyne , a. g. , shemar , s. l. , & smith , f. g. 2000 , , 315 , 534 pines , d. , & alpar , m. a. , 1985 , , 316 , 27 pizzochero , p. m. , viverit , l. , & broglia , r. a. , 1997 , , 79 , 3347 ruderman , m. 1976 , , 203 , 213 ruderman , m. 2000 , private communication sonin , e. b. 1987 , rev .
phys . , 59 , 87 wang , n. , manchester , r. n. , pace , r. t. , bailes , m. , kaspi , v. m. , stappers , b. w. & lyne , a.,g . 2000 , , 317 , 843 | the rotating neutron superfluid in the inner crust of a neutron star is threaded by quantized vortex lines .
the pinning force from lattice nuclei and the magnus force from neutron superfluid act onto a vortex line that has a finite tension .
the configuration of a vortex line in equilibrium comprises a number of pinned straight lengths separated by unpinned kinks when the rotation axis is slightly inclined to the major axis of a crystal lattice and the magnus force is not so strong .
energy of @xmath0 is required to form a kink at densities of @xmath1 .
magnus force makes the vortex line parabolic on global scale .
there exist two modes , rotational and helical ( kelvin ) , for oscillations allowed on a vortex line .
the vortex oscillations in the pinned straight segments are possible only above a minimum frequency .
we find no unstable mode that grows with time .
hence , the vortex configurations with kinks may be stable and yield the strongly pinned state .
the essence of our results may not be altered even when we consider the vortex behaviors in a polycrystalline structure .
our studies suggest that pinning of a vortex line in a polycrystalline structure would be still strong enough to explain the large glitches . |
when an ionizing particle crosses a nuclear track detector ( ntd ) sheet , it produces damages at the level of polymeric bonds around its trajectory , forming the so - called `` latent track '' . for a particle with charge @xmath5 and velocity
@xmath6 the damage depends on the ratio @xmath7 . the chemical etching in a basic solution results in the formation of etch - pit cones in both faces of the sheet .
the formation of etch - pit cones is regulated by the bulk etching rate , @xmath8 , and the track etching rate , @xmath9 , i.e. the velocities at which the undamaged and damaged materials are removed .
etch - pit cones are formed if @xmath9@xmath10@xmath8 . the response of the detector
is given by the ratio @xmath11 as a function of the restricted energy loss ( rel ) .
the poly - allyl - diglycol carbonate polymer , commercially known as cr39 , is the most sensitive ntd .
it is sensitive for a wide range of charges down to @xmath12 in the low velocity and in the relativistic regions @xcite and can reach a charge resolution @xmath13 @xcite .
cr39 was used to search for exotic particles , like magnetic monopoles and strange quark matter ( sqm ) @xcite , to study cosmic ray composition @xcite and for environmental studies @xcite . in this paper
the results of some applications of cr39 ntd are presented .
a stack composed of cr39 and makrofol sheets of size @xmath14 @xmath15 with a 1 cm thick lead target was exposed to 158 a gev pb@xmath16 ions in 1996 ; a second stack with a 1 cm thick aluminum target was exposed to 158 a gev in@xmath17 ions in 2003 . both exposures were performed at the cern - sps , at normal incidence and density of @xmath182000 ions/@xmath15 .
the detector sheets before and after the target recorded respectively the incident beam ions and the ones exiting from the target , together with their nuclear fragments .
after exposures , two cr39 sheets located after the target were etched in 6n naoh + 1% ethyl alcohol solution at 70 @xmath19c for 40 h @xcite .
the addition of ethyl alcohol in the etchant improves the etched surface quality , reduces the number of surface defects and background tracks , increases the bulk etching velocity , speeds up the reaction , but raises the detection threshold . for each cr39 sheet , the etch - pit base areas for beam ions and their fragments
were measured with the elbek automatic image analyzer system @xcite .
averages were computed from measurements made on the `` front sides '' of the detector sheets .
the peaks are well separated from @xmath0 to @xmath20 in the lead case @xcite and to @xmath21 for the indium case @xcite .
the charge resolution close to the beam peak can be improved by measuring the heights of the etch - pit cones .
the heights of 1000 etch - pit cones corresponding to nuclear fragments close to the beam peaks were measured with an accuracy of @xmath22 @xmath23 m with a leica microscope coupled to a ccd camera and a video monitor .
the corresponding distributions @xcite show that each peak is well separated from the others , and a charge can be assigned to each one .
[ fig : calib](left ) shows the average ( 2 faces ) base area distribution of etched cones in cr39 from in@xmath17 ions and their fragments . in the insert the cone height distribution for @xmath24
is shown .
for each detected nuclear fragment and for the beam ions we computed the rel and the reduced etch rate @xmath11 . fig .
[ fig : calib](right ) shows a summary of cr39 calibrations with different ions and energies .
notice that a unique calibration curve describes all the data .
the availability of ion beams at the brookhaven national laboratory ( bnl ) , usa , and at the heavy ion medical accelerator in chiba ( himac ) , japan , facilities made possible to investigate the projectile fragmentation on different targets and for different projectile energies .
the present study is focused on fe@xmath25 , si@xmath26 and c@xmath27 ion interactions in ch@xmath28 , cr39 and al targets .
stacks composed of several cr39 sheets of size @xmath14 @xmath15 and with different targets were exposed to 0.3 , 1 , 3 , 5 and 10 a gev fe@xmath25 , 1 , 3 , 5 a gev si@xmath26 ions at the bnl alternating gradient synchrotron ( ags ) and nasa space radiation laboratory ( nsrl ) . the exposures to 0.41 a gev fe@xmath25 and 0.29 a gev c@xmath27 ions were performed at himac . for more details
see @xcite .
we used three and four cr39 sheets , @xmath180.7 mm thick , placed before and after the target , respectively .
the exposures were done at normal incidence , with a density of @xmath182000 ions/@xmath15 .
after exposures , the cr39 sheets were etched in 6n naoh aqueous solution at 70 @xmath19c for 30 h. the total charge changing cross sections were determined from the survival fraction of ions using the following relation @xmath29 where @xmath30 is the nuclear mass of the target ( average nuclear mass in case of polymers : @xmath31 ) ; @xmath32 and @xmath33 are the numbers of incident ions before and after the target , respectively ; @xmath34 ( g/@xmath35 ) is the target density ; @xmath36 ( cm ) is the target thickness and @xmath37 is avogadro number .
the average track base area was computed for each reconstructed ion path by requiring the existence of signals in at least two out of three detector sheets .
the charge resolution @xmath38 was about @xmath39 .
the numbers of incident and survived beam ions were determined considering the mean area distributions of the beam peaks before and after the target and evaluating the integral of the gaussian fit of the beam peaks .
[ fig : framm ] shows the beam energy dependence of the total charge changing cross sections of ( a ) fe@xmath25 projectiles on the ch@xmath28 and al targets and ( b ) si@xmath26 projectiles on the ch@xmath28 , cr39 and al targets .
our data are almost energy independent , in agreement with the data from refs . @xcite and with the predictions of the semi - empirical formula @xcite for nuclear cross sections ( solid lines ) .
the slim ( search for light monopoles ) experiment was an array of ntds with a total area of 427 m@xmath40 exposed for 4.22 years at the chacaltaya laboratory in bolivia at 5230 m a.s.l .
the array was organized into 7410 modules , each of area @xmath41 @xmath15 .
all modules were made up of : three layers of cr39 , each 1.4 mm thick ; 3 layers of makrofol , each 0.48 mm thick ; 2 layers of lexan , each 0.25 mm thick , and one layer of aluminum absorber 1 mm thick .
the main purpose of the slim experiment was the search for intermediate mass monopoles ( imms ) with masses @xmath42 gev @xcite .
these imms , predicted by some gut and supersymmetric models , may have been produced in the early universe and could be present in the cosmic radiation @xcite .
the slim detector was also sensitive to strange quark matter nuggets @xcite and q - balls @xcite .
since imms have a constant energy loss through the stacks , the subsequent chemical etching should result in collinear etch - pit cones of equal size on both faces of each detector sheet . in order to increase the detector `` signal to noise '' ratio different etching conditions @xcite were defined .
the _ strong etching _ ( 8n koh + 1.5% ethyl alcohol at 75 @xmath19c for 30 hours ) allows better surface quality and larger post - etched cones to be obtained .
this makes etch - pits easier to detect under visual scanning .
the _ soft etching _
( 6n naoh + 1% ethyl alcohol at 70 @xmath19c for 40 hours ) allows to proceed in several etching steps and study the formation of the post - etched cones .
the analysis of a slim module started by etching the uppermost cr39 sheet using strong conditions in order to reduce the cr39 thickness from 1.4 mm to @xmath43 mm .
after the strong etching , the cr39 sheet was scanned twice , with a stereo microscope , by different operators , with a 3@xmath44 magnification optical lens , looking for any possible correspondence of etch pits on the two opposite surfaces .
the measured single scan efficiency was about 99% ; thus the double scan guarantees an efficiency of @xmath45 for finding a possible signal .
a track was defined as a `` candidate '' if the computed momentum and incident angle on the front and back sides were equal to within 20% . in this case
the lowermost cr39 layer was etched in soft etching conditions , and an accurate scan under an optical microscope with high magnification ( 500@xmath44 or 1000@xmath44 ) was performed in a square region around the candidate expected position , which included the `` coincidence '' area ( @xmath180.5 @xmath15 ) . if a two - fold coincidence was detected , the cr39 middle layer was also analyzed . since no candidates were found , the global 90% c.l .
upper limits for the flux of downgoing imms and dyons with velocities @xmath46 were computed , as shown in fig .
[ fig : limite ] .
the flux limit for @xmath47 is @xmath48 @xmath49 s@xmath50 sr@xmath50 @xcite .
the 90% c.l .
upper limits for a downgoing flux of nuclearites and charged q - balls is @xmath48 @xmath49 s@xmath50 sr@xmath50 for @xmath51 @xcite .
the nuclear track detector cr39 was calibrated with different ions of different energies
. a unique curve of @xmath52 vs rel describes all the data .
the total fragmentation cross sections for fe@xmath25 , si@xmath26 and c@xmath27 ions on polyethylene , cr39 and aluminum targets were measured using cr39 .
the total cross sections do not show any observable energy dependence and are in agreement with similar data in the literature . in the slim experiment we etched and analyzed 427 m@xmath40 of cr39 .
no candidate passed the search criteria .
the 90% c.l .
upper limits for a downgoing flux of fast ( @xmath47 ) downgoing imms are at the level of @xmath53 @xmath49 sr@xmath50 s@xmath50 .
the same limits were obtained for nuclearites , strangelets and charged q - balls with @xmath54 .
s. balestra et al .
j. c55 ( 2008 ) 57 .
m. ambrosio et al . , eur .
j. c25 ( 2002 ) 511 .
t. chiarusi et al . , rad .
40 ( 2005 ) 424 . s. manzoor et al . , nucl .
suppl . 172
( 2007 ) 92 .
s. balestra et al .
. instr . meth .
b254 ( 2007 ) 254 .
a. noll et al . , nucl . tracks rad .
meas . 15 ( 1988 ) 265 .
g. giacomelli et al . , nucl .
meth . a411 ( 1998 )
41 . c. brechtmann et al . ,
c39 ( 1989 ) 2222 .
f. flesch et al . , rad .
34 ( 2001 ) 237 . c. zeitlin et al .
c56 ( 1997 ) 388 . c. zeitlin et al .
a784 ( 2007 ) 341 . | * abstract . * the nuclear track detector cr39 was calibrated with different ions of different energies . due to the low detection threshold ( @xmath0 ) and the good charge resolution ( @xmath1 for @xmath2 with 2 measurements ) ,
the detector was used for different purposes : @xmath3 fragmentation of high and medium energy ions ; @xmath4 search for magnetic monopoles , nuclearites , strangelets and q - balls in the cosmic radiation . |
pre - main sequence stars are commonly surrounded by disks , which serve the dual purpose of funneling accretion onto the central star and transporting away angular momentum .
these disks are also the formation sites of planets and connect protostars and planets physically and chemically .
the main component of gas - rich circumstellar disks , cold h@xmath2 , can not be observed directly because of a lack of transitions at suitable energy levels .
instead , constraints on disk mass , density and temperature structures depend on observations of trace species , most commonly dust and co gas ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
to access other disk properties , such as ionization levels , x - ray and uv radiation fields , turbulent mixing , chemical age and bulk composition requires the development of additional molecular probes whose formation and destruction depend critically on one or several of these properties .
the past decade has witnessed significant progress on disk chemistry theory , resulting in large numbers of predictions on how molecular abundance structures should depend on different disk properties ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a lack of observational sensitivity has however limited detections to the simplest molecules in the millimeter / submillimeter range which characterizes the bulk of the gas disk , including co , hco@xmath8 , dco@xmath8 , cn , hcn , dcn , c@xmath2h , h@xmath2co , cs , so , ch@xmath2 , n@xmath2h@xmath8 and hc@xmath1n @xcite .
this list is expected to expand with the arrival of alma and its unprecedented sensitivity and spatial resolution , which will enable the development of a more diverse set of molecular probes . in this _
, we use alma science verification observations of hd 163296 to report the first detection of cyclopropenylidene , @xmath0-c@xmath1h@xmath2 , in a circumstellar disk . at a distance of @xmath9122 pc ( @xcite ) , this herbig ae star ( stellar mass 2.3 m@xmath10 ; spectral type a1 ; age @xmath94 myr ) harbors a large disk ( a scattered light pattern extending out to a radius of @xmath9500 au , @xcite ) with strong millimeter continuum and molecular line emission ( @xcite ) . the large disk size and strong molecular emission
make this system an ideal target to search for new molecules .
the @xmath0-c@xmath1h@xmath2 molecule is a small cyclic hydrocarbon , which was first detected in space by @xcite towards sgr b2 .
since then , @xmath0-c@xmath1h@xmath2 has been detected in a wide range of astrophysical environments , including diffuse and dense clouds , protostars , planetary nebulae and extragalactic sources ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and is one of the most abundant molecules with 3 carbon atoms in the interstellar medium @xcite . in the context of disks , @xmath0-c@xmath1h@xmath2 has some favorable characteristics that may make it a powerful molecular probe , especially when observed in combination with isotopologues and chemically related molecules : 1 .
@xmath0-c@xmath1h@xmath2 has two equivalent h nuclei , which couple to generate ortho ( nuclear spin of 1 ) and para ( nuclear spin of 0 ) states .
deviations from the statistical weights of 3 to 1 may be used to constrain the formation conditions of the molecule @xcite .
@xmath0-c@xmath1h@xmath2 is also one of two isomers ; the other is the linear molecule @xmath11-c@xmath1h@xmath2 . the isomer ratio can be used to assess the fractional ionization in clouds @xcite and may provide similar constraints on disks .
3 . the formation and destruction chemistry of @xmath0-c@xmath1h@xmath2 ( and carbon chains ) are sensitive to high energy radiation .
the vertical and radial abundance structure should provide strong constraints on the penetration depth of x - rays and uv photons @xcite .
4 . within the family of hydrocarbons ,
the sensitivity to turbulent mixing increases dramatically from c@xmath2 to c@xmath1h@xmath2 , and from c@xmath1h@xmath2 to larger hydrocarbons . observed ratios and limits have the potential to constrain transport in disks @xcite .
finally , the chemistry that produces @xmath0-c@xmath1h@xmath2 is important to constrain in its own right , since @xmath0-c@xmath1h@xmath2 shares precursors with other , more complex carbon chains , which may be an important source of carbon during planet formation . in light of these possibilities
, we aim to provide first constraints on the abundance , distribution and @xmath12 ratio of @xmath0-c@xmath1h@xmath2 in the hd 163296 disk .
hd 163296 ( ra : 17@xmath1356@xmath1421287 , dec : @xmath152157@xmath162239 ; j2000.0 ) was observed on 2012 june 9 , june 23 and july 7 as part of the alma science verification ( sv ) program in band 6 . at the time of observations , 25 12-m antennae were used
( 2 7-m aca antennae were present in the dataset but flagged ) in the array with a resulting synthesized beam size of [email protected]@xmath18 ( pa=84 ) .
the source was observed with a total integration time of 84 minutes .
the alma correlator was configured to simultaneously observe four spectral windows ( two in each sideband ) .
one of those windows ( spwid # 1 ) covered both @xmath6co and c@xmath19o @xmath20 lines ( 220.399 and 219.560 ghz , respectively ) .
the data were calibrated in the casa software package ( v3.4 ) , following the detailed calibration and imaging scripts provided by the alma science verification team .
since those scripts , along with fully calibrated measurement sets , are publicly available online , we do not repeat the details here .
the 218 ghz ( 1.37 mm ) continuum was generated from the line - free channels in the two spectral windows ( spwid # 0 and # 1 ) of the lower sideband , and the flux is determined to be [email protected] mjy , which agrees with the sma observations @xcite .
excellent agreement is also seen between the sma and alma spectra for the @xmath6co and c@xmath19o @xmath20 lines with fluxes within 10% @xcite .
all of the @xmath0-c@xmath1h@xmath2 lines are continuum - subtracted and retrieved from spwid # 0 with a channel width of 0.488 mhz ( 0.67 km s@xmath22 ) covering the total 1.875 ghz bandwidth from 216.167 to 218.042 ghz .
we have binned the lines in channels of 1 km s@xmath22 to achieve higher signal - to - noise ratios and still keep the key kinematic features of the disk .
the resulting line rms noise level is estimated to be 2.2 mjy beam@xmath22 per 1 km s@xmath22 channel in line free channels .
figure [ fig : mom ] shows that the @xmath0-c@xmath1h@xmath2 @xmath3 line is clearly detected towards hd 163296 and that the position angle of its resolved velocity field is consistent with that of @xmath6co from the same data set , which shows a clear pattern of keplerian rotation .
interestingly , the @xmath0-c@xmath1h@xmath2 @xmath3 moment map displays a double - peaked morphology , indicative of an inner hole .
the @xmath0-c@xmath1h@xmath2 @xmath23 and @xmath24 lines are only tentatively detected , insufficient to resolve the velocity field or any spatial structure . to improve the sensitivity in imaging these lines
, we applied a gaussian taper to the visibility data that corresponds to 1 fwhm in the image domain , which resulted in a beam of [email protected]@xmath18 ( pa=78 ) . because the line emission changes position with channels , using a single mask for all channels to extract spectra will lead to a degradation of the signal .
instead , we create separate masks to cover the @xmath0-c@xmath1h@xmath2 @xmath3 line emission at each channel and use these masks to extract the line spectra .
the same masks are also used for other @xmath0-c@xmath1h@xmath2 lines .
figure [ fig : spec ] shows the resulting spectra of all three @xmath0-c@xmath1h@xmath2 @xmath23 lines together with @xmath6co @xmath20 line ( with its own masks covering the @xmath6co @xmath20 emission at each channel ) . the @xmath3 line is clearly detected with the expected line shape and v@xmath25 consistent with @xmath6co .
the @xmath24 line presents a consistent spectral profile , at lower signal - to - noise ratio .
in contrast , the @xmath23 line is both narrower and clearly blue - shifted , suggestive of a different physical origin , perhaps in a disk wind .
however , both the @xmath23 and @xmath24 lines are observed at the edge of the spectral window , so this apparent asymmetry should be confirmed by independent observations before any strong conclusions are drawn .
the integrated fluxes are reported in table [ tab : line ] .
these are just below the upper limits reported in two previous failed searches of @xmath0-c@xmath1h@xmath2 in disks @xcite .
we explore the distributions of @xmath0-c@xmath1h@xmath2 in hd 163296 based on a previously developed accretion disk model with well - defined temperature and density structures @xcite . to simulate and compare with the data , we assume the disk material orbits the central star in keplerian motion , and fix the disk geometric and kinematic parameters that affect the observed spatio - kinematic behavior of the disk .
the details of the stellar and accretion properties and the disk geometric and kinematic parameters are summarized in table 3 of @xcite .
we have developed a physically self - consistent accretion disk model with an exponentially tapered edge that matches the spectral energy distribution and spatially resolved millimeter dust continuum emission .
the disk temperature and density structures are further constrained by multiple co and co isotopologue line observations .
such analysis provides the essential framework for constraining the distribution of molecular abundances in protoplanetary disks . within this model framework ,
we assume that @xmath0-c@xmath1h@xmath2 is present with a constant abundance in a layer with boundaries toward the midplane and toward the surface of the disk @xcite . this assumption is motivated by chemical models ( e.g. * ? ? ?
* ) that predict a three - layered structure where most molecules are photodissociated in the surface layer , frozen out in the midplane , and have an abundance that peaks at intermediate disk heights .
the surface ( @xmath26 ) and midplane ( @xmath27 ) boundaries are presented in terms of @xmath28 , where @xmath29 is the hydrogen column density measured from the disk surface in the adopted physical model .
this simple model approach approximates the vertical location where @xmath0-c@xmath1h@xmath2 is most abundant . due to the limited signal - to - noise of the multiple @xmath0-c@xmath1h@xmath2 line detections
, we can not constrain the location of this vertical layer based on the @xmath0-c@xmath1h@xmath2 data .
we therefore fix the vertical surface boundary @xmath26 as 0.32 ( @xmath30 ) and the midplane boundary @xmath27 as 3.2 ( @xmath31 ) , between which lies the expected location of most molecules with peak abundances in the warm molecular layer @xcite .
we model the radial column density distribution of @xmath0-c@xmath1h@xmath2 as a power law n@xmath32 , where n@xmath33 is the column density at 100 au in @xmath7 , @xmath34 is the distance from the star in au , and @xmath35 is the power - law index . since the emission shows a clear indication of a central hole , we fit for both an inner radius r@xmath36 and outer radius r@xmath37 together with the power - law parameters ( n@xmath33 and @xmath35 ) .
using the structure model and the fixed vertical distributions , we compute a grid of synthetic @xmath0-c@xmath1h@xmath2 j=@xmath3 visibility datasets over a range of r@xmath37 , r@xmath36 , @xmath35 and n@xmath33 values and compare with the observations .
the best - fit model is obtained by minimizing @xmath38 , the weighted difference between the real and imaginary part of the complex visibility measured in the ( @xmath39)-plane sampled by the alma observations .
we use the two - dimensional monte carlo model ratran @xcite to calculate the radiative transfer and molecular excitation .
the collisional rates are taken from @xcite and the molecular data file is retrieved from the leiden atomic and molecular database @xcite .
figure [ fig : chi2 ] shows the @xmath38 surfaces for the r@xmath36 and @xmath40 versus the power law index @xmath35 .
we find the @xmath35 is constrained between -2.5 and -1.5 , r@xmath36 1540 au and r@xmath37 150200 au ( within 1@xmath41 ) with best - fit values at -2 , 30 and 165 au , respectively ( reduced @xmath42 ) . by
fixing the above parameters at their best - fit values , n(@xmath0-c@xmath1h@xmath2 ) is determined to be [email protected] @xmath1710@xmath5 @xmath7 at 100 au .
figure [ fig : chmap ] presents comparisons between the observed channel maps and the best - fit model .
the residual image does nt show any significant emission , which indicates that the @xmath0-c@xmath1h@xmath2 emission is consistent with that arising from a keplerian rotating disk .
the @xmath3 line is actually blended with the @xmath43 ( ortho ) and @xmath44 ( para ) lines at the same frequency . since the einstein @xmath45-coefficients for both ortho and
para lines are the same , 5.393@xmath1710@xmath46 s@xmath22 @xcite , we fit the data with the ortho line only and the resulting column density can be accounted as the sum of both ortho and para forms of the molecule . for the weaker @xmath24 ortho line ,
the signal - to - noise ratio is too low to provide any constraints on the @xmath0-c@xmath1h@xmath2 the distribution . by assuming the @xmath0-c@xmath1h@xmath2 distribution derived from the @xmath3 line , however
, we can use the unresolved line flux to constrain the @xmath12 ratio .
the resulting best - fit n(c@xmath1h@xmath2-(@xmath47))[email protected] @xmath1710@xmath5 and n(c@xmath1h@xmath2-(@xmath48)[email protected] @xmath1710@xmath5 @xmath7 at 100 au ( as shown above ) .
hence @[email protected] , which is close to the statistical value of 3 , but the large uncertainty implies that better data are required to use this ratio as a probe of formation conditions .
we have clearly detected @xmath0-c@xmath1h@xmath2 in the disk around hd 163296 .
this is one of the two largest molecules detected in disks so far , the other being hc@xmath1n , also a carbon - chain molecule , detected recently using deep single - dish observations @xcite . using the spatially and
spectrally resolved line emission , we have constrained the radial distribution of @xmath0-c@xmath1h@xmath2 in the hd 163296 disk as a ring structure ranging from @xmath930 to @xmath9165 au .
the outer edge of @xmath49250 au ( 2@xmath41 limit ) is much smaller than the gas disk size which extends to 500 au @xcite , which suggests that @xmath0-c@xmath1h@xmath2 formation is slow or that @xmath0-c@xmath1h@xmath2 destruction is fast in the outer disk . in terms of formation
, the best - fit outer radius coincides with the previously determined location of the co midplane snow line , and it is possible that co freeze - out limits the carbon available in the gas phase to form hydrocarbons in general and @xmath0-c@xmath1h@xmath2 in particular .
a second characteristic of the outer disk is that it is rather tenuous because of a rapidly decreasing column density outside of 200 au @xcite .
this may result in efficient penetration of radiation and thus destruction of the easily dissociated @xmath0-c@xmath1h@xmath2 .
explicit model predictions are required to resolve which effect drives the disappearance of @xmath0-c@xmath1h@xmath2 in the outer disk .
no existing disk chemistry model in the literature contains predictions for the radial distribution of @xmath0-c@xmath1h@xmath2 .
based on strong correlations found between @xmath0-c@xmath1h@xmath2 and cch in several astrophysical environments @xcite , predictions on cch in disks can be used to shed further light on the origins of @xmath0-c@xmath1h@xmath2 .
@xcite predicts a cch central cavity in the absence of x - ray radiation ; when x - rays are present , and hd 163296 is a known x - ray emitter @xcite , the cch column is instead centrally peaked . despite the presence of x - rays , the central cavity could be due to a uv dominated radiation field .
this may be tested by observing @xmath0-c@xmath1h@xmath2 towards t tauri stars with weak uv fields and strong x - rays , and towards additional herbig ae stars with weak x - rays .
the inner cavity could also be a product of the main formation pathway of @xmath0-c@xmath1h@xmath2 .
carbon chains can form through at least three different pathways : ( 1 ) ion - neutral reactions at low temperatures , ( 2 ) photo - erosion of larger carbonaceous compounds , and ( 3 ) neutral - neutral reactions at luke - warm temperatures following ch@xmath50 ice evaporation at @xmath930 k @xcite .
detailed modeling of @xmath0-c@xmath1h@xmath2 radial distributions in each of these formation scenarios are clearly needed to use @xmath0-c@xmath1h@xmath2 rings as tracers of the radiation field and other disk characteristics .
in addition to more detailed modeling , additional observations are needed to constrain whether different hydrocarbons , especially cch and @xmath0-c@xmath1h@xmath2 , are correlated in disk environments , and to constrain the vertical distribution of these molecules .
the latter requires multiple lines with a range of excitation energies . with the combined advance in theory and observations
, @xmath0-c@xmath1h@xmath2 may become one of the more useful probes of penetration of the radiation fields in disks .
this paper makes use of the following alma data : ads / jao.alma#2011.0.00010.sv .
alma is a partnership of eso ( representing its member states ) , nsf ( usa ) and nins ( japan ) , together with nrc ( canada ) and nsc and asiaa ( taiwan ) , in cooperation with the republic of chile .
the joint alma observatory is operated by eso , aui / nrao and naoj . + we would like to thank an anonymous referee for thoughtful suggestions for the paper .
we also acknowledge nasa origins of solar systems grant no .
nnx11ak63 . , k. m. , carilli , c. l. , & reid , m. j. 1999 , in astronomical society of the pacific conference series , vol .
156 , highly redshifted radio lines , ed . c. l. carilli ,
s. j. e. radford , k. m. menten , & g. i. langston , 218 | we report the first detection of @xmath0-c@xmath1h@xmath2 in a circumstellar disk . the @xmath0-c@xmath1h@xmath2 j=@xmath3 line ( 217.882 ghz ) is detected and imaged through atacama large millimeter array ( alma ) science verification observations toward the disk around the herbig ae star hd 163296 at 0.8@xmath4 resolution .
the emission is consistent with that arising from a keplerian rotating disk .
two additional @xmath0-c@xmath1h@xmath2 transitions are also tentatively detected , bolstering the identification of this species , but with insufficient signal - to - noise ratio to constrain the spatial distribution . using a previously developed model for the physical structure of this disk
, we fit a radial power - law distribution model to the @xmath0-c@xmath1h@xmath2 @xmath3 emission and find that @xmath0-c@xmath1h@xmath2 is present in a ring structure from an inner radius of about 30 au to an outer radius of about 165 au .
the column density is estimated to be 10@xmath510@xmath6 @xmath7 .
the clear detection and intriguing ring structure suggest that @xmath0-c@xmath1h@xmath2 has the potential to become a useful probe of radiation penetration in disks . |
in large systems , macroscopic superpositions of quantum states have extremely small decoherence times , making them impossible to observe @xcite .
the largest superpositions generated so far are superpositions of coherent states ( css ) of a cavity field , and their progressive transformation into statistical mixtures as time evolves has been observed @xcite . the first proposal to generate a macroscopic superposition of css ( mscss ) with light consisted of sending photons through a medium presenting a strong kerr nonlinearity @xcite . in such media ,
the dynamical phases of fock states are nonlinear in the photon number , thus the phase of an initial cs is split . in metastable vapors of ultracold bosonic atoms ,
interactions between atoms lead to similar nonlinearities , whose strength can be tuned experimentally by using feshbach resonances @xcite . by trapping optically the condensed atoms in a double - well potential one realizes an external bose - josephson junction ( bjj ) ; an internal bjj is obtained by trapping in a single well atoms in two distinct hyperfine states coupled by a resonant field . in analogy with light ,
the nonlinear dynamics generates mscss after a sudden quench to zero of the tunnel amplitude ( for external bjjs ) or a switch off of the coupling field ( for internal bjjs ) @xcite .
whereas mscss of light are presumably destroyed after a single photon loss @xcite , the situation is less clear for atoms . up to now only squeezed states which are produced at earlier times than mscss
have been observed in bjjs with a few hundreds of atoms @xcite . in order to know if future experiments could produce mscss in bjjs , it is desirable to study how robust they are with respect to decoherence .
decoherence effects due to atom losses and phase noise on squeezed states have been analyzed in detail @xcite .
for what concerns mscss , only decoherence caused by noise due to photon scattering @xcite and phase noise @xcite has been studied so far . under current experimental conditions
the first noise is negligible and the second one can be reduced by using spin echo @xcite ; in contrast , atom losses are unavoidable . in this article
, we investigate whether a macroscopic superposition can be formed in a bjj even in the presence of atom losses .
we are primarily interested in internal bose - josephson junctions , such as those studied experimentally in @xcite .
we focus on two - body losses , due to scattering of two atoms in the magnetic trap which changes their spin and gives them enough kinetic energy to be ejected from the trap .
these loss processes are particularly detrimental in the experiments of refs . @xcite . our analysis , however , also applies to one- and three - body losses @xcite .
we first analyze the dynamics of a lossy bjj from the point of view of state conditioning , assuming that the total number @xmath0 of condensed atoms can be measured precisely , both initially and at the mscss formation time @xmath1 .
we study how much coherence is destroyed by a single loss event occurring at a random time between @xmath2 and @xmath1 .
we find quite different answers depending on the degree of asymmetry between the loss rates and interaction energies in the two modes of the junction .
finally , we show that for strongly asymmetric losses one can protect the coherence of the mscss by suitably tuning the interaction energies , even after many loss events and in the absence of measurement of the atom number .
we consider an internal bjj in the quantum regime .
initially , @xmath3 atoms are all in the same single - particle state which is a symmetric superposition of the two internal states .
this corresponds to the ground state in the regime where tunneling dominates interactions , described by a spin coherent state @xmath4^{n_0/2 } } { | n_1,n_0-n_1 \rangle}\end{aligned}\ ] ] with @xmath5 and @xmath6 . here
@xmath7 is the joint eigenstate of the number operators @xmath8 in the mode @xmath9 ( fock state ) .
the dynamics following a sudden quench of the coupling ( tunnel energy ) to zero is given by the two - mode bose - hubbard hamiltonian @xcite @xmath10 where @xmath11 and @xmath12 are the internal energy and the interaction energy between two atoms in the same mode @xmath13 , respectively , and @xmath14 is the inter - mode interaction energy . setting @xmath15 , the interactions in eq .
( [ eq - h_0 ] ) sum up to a non - linear term @xmath16 , with @xmath17 .
the time - evolved state at time @xmath18 is a superposition of css , @xmath19 , with @xmath20 and @xmath21 @xcite . in the presence of two - body losses the markovian master equation for the density matrix @xmath22 of the condensed atoms reads @xcite ( setting @xmath23 ) @xmath24 + \sum_{i=1,2 } \gamma_i \bigl ( { \hat{a}_i}^2 \ , \hat{\rho } \ ,
( { { \hat{a}_i}}^2)^\dagger \\ { \nonumber } & & - \frac{1}{2 } \bigl\ { { \hat{n}_i } ( { \hat{n}_i}-1 ) , \hat{\rho } \bigr\ } \bigr ) + \gamma_{12 } \bigl ( { \hat{a}_1}{\hat{a}_2}\ , \hat{\rho } \ , { \hat{a}_1}^\dagger { \hat{a}_2}^\dagger - \frac{1}{2 } \bigl\ { { \hat{n}_1}{\hat{n}_2 } , \hat{\rho } \bigr\ } \bigr)\end{aligned}\ ] ] where @xmath25 and @xmath26 are the loss rates of two atoms in the same mode @xmath13 and of one atom in each mode , respectively , and @xmath27 denotes the anticommutator . since eq . ( [ eq - master_equation ] ) does not couple subspaces with different total atom numbers @xmath28 and @xmath29 initially , at all times @xmath30 has the block structure @xmath31 , where @xmath32 is the probability to have @xmath33 atoms at time @xmath34 and @xmath35 is the corresponding conditional state .
this decomposition is naturally accounted for by quantum trajectories @xmath36 , @xmath37 being the unnormalized wave function when @xmath38 loss events occur at times @xmath39 in the channels @xmath40 @xcite , @xmath41 for two - body losses one has three loss channels @xmath42 , @xmath43 , and @xmath44 .
the corresponding jump operators are @xmath45 , @xmath46 , and @xmath47 .
the dynamics between loss events is given by the effective hamiltonian @xmath48 .
the damping operator @xmath49 describes the gain of information on the system resulting from the knowledge that no loss occurred .
the conditional state @xmath50 after a detection of @xmath51 atoms at time @xmath34 is obtained by averaging @xmath52 over the @xmath38 jump times @xmath53 and channels @xmath54 , @xmath55 where @xmath56 is the joint distribution of the @xmath53 , @xmath54 , and @xmath38 @xcite . by further summing over the number of jumps @xmath38
one gets the total density matrix @xmath57 which is a solution of eq .
( [ eq - master_equation ] ) . in order to understand the effect of losses we analyze separately each @xmath33-atom sectors .
when no loss occurs in the time interval @xmath58 $ ] , from eqs .
( [ eq - no_jump ] ) and ( [ eq - d ] ) we obtain the unnormalized conditional state @xmath59 in the fock basis , @xmath60 } { \langle n_1,n_2 | } \hat{\rho}^{(0 ) } ( t ) { | n_1',n_2 ' \rangle}\,,\end{aligned}\ ] ] where @xmath61 is the lossless density matrix and @xmath62 up to an irrelevant constant , with @xmath63/(\gamma_1+\gamma_2-\gamma_{12})$ ] . for @xmath64
the matrix elements of @xmath65 have gaussian moduli peaked at @xmath66 with a width @xmath67 .
for symmetric loss rates @xmath68 and @xmath69 , this peak coincides with the center of the gaussian damping factor in eq .
( [ eq - state_no_jump ] ) .
thus the mscss formed at time @xmath1 is affected by damping when @xmath70 .
in contrast , for @xmath71 the damping factor is centered at @xmath72 and its effect on the mscss sets in at the much smaller rate @xmath73 .
we focus now on trajectories having one loss event in channel @xmath75 at the random time @xmath76 $ ] .
the instantaneous jump transforms a cs into a cs , @xmath77 .
this cs is rotated on the bloch sphere by the evolution under the nonlinear effective hamiltonian @xmath78 due to changes in energy and damping before and after the jump , yielding ( see appendix [ sec - app1 ] ) @xmath79 where the random angles @xmath80 and @xmath81 depend on the random loss time @xmath82 , the interaction energies @xmath83 , @xmath84 , @xmath85 , and the loss rate differences @xmath86 , @xmath87 , and @xmath88 . hence , apart from reducing @xmath33 and producing damping , atom losses are identical to external @xmath89 and @xmath90 noises rotating the state around the @xmath91 axis by a complex angle @xmath92 @xcite .
these noises have fluctuations @xmath93 and @xmath94 , where @xmath95 is the loss - time fluctuation and we assumed @xmath96 .
( color online ) modulus of the density matrix @xmath97 in the subspace with @xmath98 atoms in the fock basis at time @xmath99 , obtained by an exact diagonalization of eq .
( [ eq - master_equation ] ) .
upper panels : symmetric losses rates ( @xmath100 ) ; lower panels : asymmetric losses ( @xmath101 , @xmath102 ) . left column : symmetric energies ( @xmath103 ) ; right column : asymmetric energies ( @xmath104 , @xmath105 ) .
other parameters : @xmath106 , @xmath107 .
, title="fig : " ] ( color online ) modulus of the density matrix @xmath97 in the subspace with @xmath98 atoms in the fock basis at time @xmath99 , obtained by an exact diagonalization of eq .
( [ eq - master_equation ] ) .
upper panels : symmetric losses rates ( @xmath100 ) ; lower panels : asymmetric losses ( @xmath101 , @xmath102 ) . left column : symmetric energies ( @xmath103 ) ; right column : asymmetric energies ( @xmath104 , @xmath105 ) .
other parameters : @xmath106 , @xmath107 .
, title="fig : " ] ( color online ) modulus of the density matrix @xmath97 in the subspace with @xmath98 atoms in the fock basis at time @xmath99 , obtained by an exact diagonalization of eq . ( [ eq - master_equation ] ) .
upper panels : symmetric losses rates ( @xmath100 ) ; lower panels : asymmetric losses ( @xmath101 , @xmath102 ) . left column : symmetric energies ( @xmath103 ) ; right column : asymmetric energies ( @xmath104 , @xmath105 ) .
other parameters : @xmath106 , @xmath107 .
, title="fig : " ] ( color online ) modulus of the density matrix @xmath97 in the subspace with @xmath98 atoms in the fock basis at time @xmath99 , obtained by an exact diagonalization of eq .
( [ eq - master_equation ] ) .
upper panels : symmetric losses rates ( @xmath100 ) ; lower panels : asymmetric losses ( @xmath101 , @xmath102 ) . left column : symmetric energies ( @xmath103 ) ; right column : asymmetric energies ( @xmath104 , @xmath105 ) .
other parameters : @xmath106 , @xmath107 .
, title="fig : " ] let us concentrate on the mscs formation time @xmath108 and take @xmath69 .
we first consider _ weak losses @xmath109 _ and _ symmetric energies @xmath110_. in this regime the @xmath89-noise is negligible since @xmath111 is much smaller than the quantum fluctuations @xmath112 of a cs .
in contrast , one has large @xmath90 fluctuations @xmath113 in the two loss channels @xmath114 , equal to the inter - component phase separation of the mscss ( we use here @xmath115 and @xmath116 ) .
such a strong @xmath90 noise has a relatively small effect on the coherences , albeit it transforms the corresponding statistical mixture of css into a mixture of fock states ( phase relaxation ) @xcite .
actually , in the absence of losses in the second mode ( @xmath117 ) , the off - diagonal elements of the conditional density matrix @xmath118 are weakly affected by losses for @xmath119 modulo @xmath120 , as shown in appendix [ sec - app3 ] and in fig .
[ fig2 ] ( lower left panel ) .
the global shift to values @xmath121 is due to the gaussian damping caused by @xmath78 in eq .
( [ eq - wave function_one_jump_bis_bis ] ) . for @xmath68 , instead
, @xmath122 is almost diagonal in the fock basis , see the upper left panel in fig.[fig2 ] .
this is due to a cancellation ( occurring only for @xmath123 ) when summing the contributions of the two channels ( see appendix [ sec - app3 ] ) .
we now consider _ asymmetric energies @xmath124 _ , still assuming weak losses @xmath109 . in order to keep @xmath125 constant , we vary @xmath126 , @xmath127 while fixing @xmath128 .
interestingly , it is possible to protect one channel , say @xmath129 , against @xmath90 noise by choosing @xmath130 and @xmath131 , at the expense of enlarging noise in the other channel
. then @xmath132 and @xmath133 .
if only the first channel loses atoms , the conditional state after a single loss event is then close to a mscss with @xmath98 atoms , apart from the damping described in eq .
( [ eq - state_no_jump ] ) , as seen in the lower right panel in fig .
[ fig2 ] . for symmetric losses
, @xmath134 has also large off - diagonal elements in the fock basis due to the probability @xmath135 of losing atoms in the protected channel ( fig .
[ fig2 ] , upper right panel ) .
hence , the mscss can be protected by tuning the interaction energies so that @xmath136 in the mode @xmath13 with the highest loss rate @xmath25 . for @xmath137rb atoms used in the experiment of ref .
@xcite , a magnetic field in resonance with one of the feshbach peaks for @xmath129 , @xmath43 , or @xmath44 must be applied in order to have a nonzero @xmath138 ( actually , without magnetic field one has nearly @xmath139 ) . since
two - body losses are mostly important in the upper internal level @xmath140 , in order to better preserve the coherence , @xmath141 must be tuned such that @xmath142 .
let us remark that , although our results also apply to external bjjs , for such bjjs @xmath143 and thus one must tune the interaction energy @xmath12 to zero to switch - off phase noise in the well @xmath13 .
but the loss rate @xmath25 depends on @xmath12 and this tuning actually decreases @xmath25 , so that the protection of the mscss is a trivial effect . in contrast , for internal bjjs choosing @xmath144 does not decrease the loss rates , but it diminishes decoherence in the loss channel @xmath13 at weak losses .
let us now turn to the _ intermediate loss rate _ regime .
the @xmath90 noise decreases when increasing @xmath145 since the loss - time fluctuations @xmath146 decrease .
indeed , we find @xmath147 ( @xmath9 ) for @xmath64 .
physically , at increasing @xmath145 the loss has more chance to occur at small times , while for small @xmath145 , @xmath82 is equally distributed in @xmath58 $ ] .
note , however , that the probability @xmath148 of losing only two atoms decreases by increasing @xmath145 .
the wavefunction @xmath150 after @xmath151 jumps is still given by eq .
( [ eq - wave function_one_jump_bis_bis ] ) upon replacing @xmath98 by @xmath152 and the angles of the cs by @xmath153 and @xmath154 with @xmath155 where @xmath156 and @xmath157 are the angles corresponding to the single loss event .
thus the aforementioned effects persist . for weak symmetric losses , though , keeping coherence by switching off phase noise in one channel is harder since the probability that all jumps occur in that channel decreases exponentially with @xmath38 .
this means that for many loss events our proposal for protecting mscss is only efficient for strongly asymmetric loss rates .
we now show that for strongly asymmetric losses not only the conditional states but also the quantum correlations in the full density matrix @xmath158 can be protected by tuning the interactions .
we measure the amount of quantum correlations with the quantum fisher information @xmath159 , related to the best achievable phase precision in a mach - zehnder interferometer using @xmath160 as input state by @xmath161 @xcite .
hence @xmath162 implies phase accuracy beyond the shot noise limit @xmath163 @xcite . in our case @xmath164 = \sum_{n=0}^{n_0 }
w_n(t ) f_n ( t)$ ] where @xmath165 is the fisher information of the conditional state @xmath166 in the subspace with @xmath33 atoms .
the latter is given by @xmath167 where @xmath168 and @xmath169 is the angular momentum operator in the direction @xmath170 .
we optimize @xmath171 over all directions @xmath170 of the interferometer . in a lossless bjj , the two - component superposition has the highest possible value @xmath172 , nearly twice larger than that of mscss with @xmath173 components @xcite .
( color online ) fisher information in the subspace with @xmath98 atoms as a function of the loss rate @xmath174 ( in semilogarithmic scale ) at time @xmath175 , with @xmath176 .
solid black line : @xmath177 , @xmath103 ; dot - dashed brown line : @xmath177 , @xmath178 , @xmath179 ; dotted red line : @xmath180 , @xmath181 ; dashed blue line : @xmath182 , @xmath183 .
the optimization over the interferometer directions is done independently in the @xmath74 subspace .
insets : husimi functions for the values of @xmath25 and @xmath184 corresponding to the circles on the curves .
other parameters as in fig .
[ fig2 ] . ]
figure [ fig3 ] shows the fisher information @xmath185 in the subspace with @xmath186 atoms , corresponding to the state conditioned to a single loss event . for symmetric energies ,
the low values of @xmath187 at small symmetric loss rates are direct consequences of the cancellation among channels ( see appendix [ sec - app3 ] ) . for @xmath102
much larger values are found , showing that the aforementioned non - vanishing inter - component coherences carry useful quantum correlations . at intermediate rates , @xmath188 increases and
reaches a maximum as a result of the reduced phase noise when @xmath68 ( note that this peak will not be seen on @xmath189 because of the rapid decay of the probability @xmath190 by increasing @xmath145 ) .
this reduction is clearly seen on the husimi distributions @xmath191 which display a flat profile for small losses and two emerging peaks for larger losses ( insets in fig .
[ fig3 ] ) . for asymmetric losses @xmath102 , by choosing @xmath178 we have @xmath192 in the small loss limit and the husimi function has two peaks at the cs phases @xmath193 , in agreement with our prediction that the conditional state converges to a two - component superposition
; at intermediate rates @xmath188 decreases with @xmath174 because the gaussian damping compensates phase noise reduction . at large losses
this damping transforms @xmath97 into a superposition of fock states with @xmath194 or @xmath195 atoms in the first mode @xcite . for symmetric losses and even @xmath3 , @xmath97
is transformed instead into the fock state @xmath196 having a larger fisher information @xmath197 . our main result is presented in fig .
[ fig1 ] , which displays the total fisher information @xmath198 obtained from an exact diagonalization of eq .
( [ eq - master_equation ] ) for various rates and energies , keeping the same value of @xmath138 and of the mean number of atoms at time @xmath99 ( see the inset ) .
we find that for @xmath102 , tuning the energies has a strong effect on @xmath198 at times @xmath199 , yielding to larger fisher informations than for squeezed states .
quantum correlations in the total density matrix @xmath200 are then preserved even after the loss of @xmath201 of atoms .
( color online ) total quantum fisher information @xmath171 vs time @xmath34 ( in units of @xmath202 ) . from top to bottom : @xmath203 , @xmath182 , @xmath204 ( blue dashed line ) ; @xmath205 , @xmath102 , @xmath206 ( red dotted line ) ; @xmath207 , @xmath178 , @xmath179 ( brown dot - dashed line ) ; @xmath208 , @xmath209 ( black solid line ) .
other parameters as in fig .
[ fig2 ] . dashed horizontal line : @xmath210 for the highest squeezed state in the lossless case .
inset : average number of atoms vs time for the same parameters . ]
we have shown that the interplay of atom losses and interactions in bjjs leads to different decoherence scenarios even after a single loss event .
in particular , for strongly asymmetric two - body losses one can protect superpositions by tuning the interaction energies , leading to useful states for high - precision atom interferometry applications .
such asymmetric losses occur in internal bjjs with @xmath137rb atoms in the hyperfine states @xmath211 and @xmath212 , subject respectively to fast two - body and slow one - body losses @xcite .
finally , we note that the @xmath90 noise associated to the loss times @xmath53 could be used to obtain an indirect measurement of @xmath53 from the interferometric estimation of the phase rotation of the mscss , yielding to an observation of quantum jumps like in cavity qed @xcite .
we are grateful to k. rzaewski , f. hekking , p. treutlein , c. gross , and m. oberthaler for helpful discussions . d.s and k.p acknowledge support from the project anr-09-blan-0098 - 01 , a.m. from the erc handy - q project , and k.p . from the polish government funds for the years 2010 - 2012 and the project `` decoherence in long range interacting quantum systems and devices '' sponsored by the baden - wrttemberg stiftung .
let us show that the quantum trajectory @xmath213 having one loss event in channel @xmath75 at the random time @xmath76 $ ] is given by eq .
( [ eq - wave function_one_jump_bis_bis ] ) .
we prove this formula for @xmath129 ( the other cases @xmath140 and @xmath214 are similar ) .
we first determine how an initial fock state @xmath7 is transformed if two atoms are lost in mode @xmath195 at time @xmath82 .
using eq .
( [ eq - no_jump ] ) , this state becomes @xmath215 at time @xmath34 , where @xmath216 is a complex dynamical phase and @xmath217 are the quadratic eigenvalues of @xmath78 . setting @xmath218 yields @xmath219 up to @xmath220-independent constants , with @xmath221 and @xmath222 .
the two last terms in ( [ eq - a1 ] ) correspond respectively to the energy and damping changes due to the atom loss at time @xmath82 . replacing @xmath7 in the fock - state expansion of the initial cs
[ see eq .
( [ eq - def_coherent_state ] ) ] by the above transformed state , we get eq .
( [ eq - wave function_one_jump_bis_bis ] ) .
we now determine the conditional density matrix @xmath223 for a single loss event ( @xmath224 ) in the fock basis . using eq .
( [ eq - no_jump ] ) we find the distribution @xmath225^{n_0 - 2 } \\ { \nonumber } & & \hspace*{5 mm } \times \| e^{-i t { \hat{h}_{\rm eff } } } { | n_0 - 2;\theta_m(s),0 \rangle}\|^2 \frac{n_0 ( n_0 - 1)}{4}\end{aligned}\ ] ] with @xmath226 , @xmath9 , and @xmath227 .
averaging over trajectories as in eq .
( [ eq - density_matrix_subspace_n_0-j ] ) yields @xmath228 where @xmath229 is the conditional state having no loss in @xmath58 $ ] for an initial cs with @xmath98 atoms , and @xmath230}}{g_m\!+\ !
\delta_m ( n+n'\!-\!n_0\!+\!2 ) \ ! + \ ! 2
{ { \rm{i}}}\chi_m ( n - n')}.\ ] ] equation ( [ eq - density_matrix_n_0 - 2_sector_in_fock_basis ] ) shows that the density matrix conditioned to a single loss event is a superposition of css with @xmath98 atoms modulated by the envelope @xmath231 and the damping factor of eq .
( [ eq - state_no_jump ] ) .
the above calculation can be generalized to trajectories with @xmath151 jumps between @xmath2 and @xmath34 . for @xmath232 and @xmath233 , @xmath234
is still given by eq .
( [ eq - density_matrix_n_0 - 2_sector_in_fock_basis ] ) upon replacing @xmath235 by @xmath236^j$ ] .
in this appendix we study the density matrix ( [ eq - density_matrix_n_0 - 2_sector_in_fock_basis ] ) in the subspace with @xmath186 atoms in the weak loss regime @xmath237
. one can show @xcite that the density matrix @xmath238 associated with the superposition of css @xmath239 formed at time @xmath1 in the absence of losses has the following structure in the fock basis : @xmath240 is the sum of two matrices , the first one having zero matrix elements for @xmath241 modulo @xmath120 and corresponding to the statistical mixture of css @xmath242 , and the second one having zero matrix elements for @xmath243 modulo @xmath120 and corresponding to the coherences @xmath244 . for weak losses @xmath237 and equal interaction energies @xmath110
, @xmath245 vanishes at the values @xmath246 ( @xmath247 ) corresponding to the nonzero off - diagonal elements of the aforementioned statistical mixture of css , and decays like @xmath248 for the other off - diagonal elements encoding the coherences .
hence the statistical mixture of css is transformed after one loss event into a mixture of fock states ( complete phase relaxation ) .
the inter - component coherences are , however , non vanishing , except for @xmath68 and @xmath123 , because then the contributions of the two channels cancel each other , @xmath249 for @xmath250 .
this explains the low and high values of @xmath251 for symmetric and asymmetric losses seen in fig . 2 ( recall that the coherences are responsible for the quantum correlations quantified by the fisher information ) . taking now asymmetric interaction energies , @xmath178 and @xmath252 , one infers from @xmath253 that @xmath254 when @xmath102 .
therefore , the conditional state is a two - component superposition slightly modified by gaussian damping . | we study how macroscopic superpositions of coherent states produced by the nondissipative dynamics of binary mixtures of ultracold atoms are affected by atom losses .
we identify different decoherence scenarios for symmetric or asymmetric loss rates and interaction energies in the two modes . in the symmetric case the quantum coherence in the superposition
is lost after a single loss event . by tuning appropriately the energies
we show that the superposition can be protected , leading to quantum correlations useful for atom interferometry even after many loss events . |
neutral gas at high redshifts is easiest to detect in absorption against bright background sources .
not surprisingly , most of what we know about the content and evolution of neutral gas in the universe comes from the study of absorption lines seen in the spectra of distant qsos .
the number density ( per unit redshift ) of these absorption lines is a strong function of the column density , with low column density ( @xmath7 atoms @xmath8 ) systems being several orders of magnitude more common than high column density ( @xmath9 atoms @xmath8 ) systems . nonetheless , the bulk of the neutral gas at high redshift is contained in these rare high hi column density absorbers .
it is principally for this reason that these objects ( called damped lyman-@xmath1 systems or dlas ) are obvious candidates for the precursors of today s galaxies .
further , the gas mass in dlas at @xmath10 is comparable to the stellar mass in galaxies at @xmath11 ( e.g. @xcite ) , consistent with the idea that the absorbers have converted their gas into stars in the intervening period . understanding the nature of dlas at different redshifts is clearly important in tracing the evolution of galaxies ; for this reason , the absorbers have been the subject of considerable study over the last two decades .
unfortunately , since qsos are point sources , optical absorption studies alone are unable to constrain the transverse size of the absorbers .
the typical size and mass of dlas have hence long been controversial issues .
traditionally , dlas have been modelled as large proto - spirals ( @xcite ) .
some support for this model comes from the shapes and widths of the absorption profiles produced by ions such as siii ( which are associated with neutral hi ) .
the large velocity widths ( @xmath12 km s@xmath13 ) and the asymmetric shapes of these lines have been successfully modelled as arising from gas in a thick spinning disk ( @xcite , 1998 ) . however , models involving infall or random motions of smaller gas clouds have also been found to succesfully reproduce the observed velocity profiles ( @xcite ) . at low redshifts ,
the galaxies in which the damped absorption arises can be directly studied by ground - based or hst observations .
contrary to expectations , low redshift dlas appear to be associated with a wide variety of galaxy types , including dwarf and low surface brightness ( lsb ) galaxies ( e.g. @xcite ) and not exclusively ( or even predominantly ) with spiral galaxies .
besides this , the majority of dlas tend to have low metallicities ( @xmath14 solar ) at all redshifts , with very little evolution in their metallicity with redshift ( @xcite ) ; further , they also do not show the expected @xmath1/fe enrichment pattern expected for spiral galaxies ( @xcite ) , suggesting a different star formation history than that of spirals . finally , 21 cm absorption studies ( @xcite ) have shown that the majority of dlas have high spin temperatures ( @xmath15 k ) , far higher than those observed in the milky way or local spirals ( @xmath16 k ) .
all these results indicate that damped absorption is likely to originate in _ all _ types of galaxies and not merely in luminous disk systems .
of course , the above results do not preclude the possibility that the galaxies responsible for the damped absorption are indeed massive gas - rich disks , but have not undergone much star formation and hence have both low luminosities as well as low metallicities .
such low surface brightness systems could well have a larger fraction of the warm phase of neutral hi and hence , a high spin temperature ( @xcite ) .
however , for systems at low redshift , 21 cm emission studies can be used to get direct estimates of the hi mass of the absorbers ; one can thus directly test the above hypothesis , that dlas are massive , low luminosity galaxies .
deep searches for hi emission from two nearby dlas have resulted in non - detections ( @xcite ) . in both cases ,
the 3 @xmath17 upper limit to the hi mass is @xmath18 1/3 the hi mass of an l@xmath19 spiral .
thus , in these two cases at least , the absorption does not arise in an optically faint , but extremely gas rich galaxy . in this paper
, we discuss giant metrewave radio telescope ( gmrt ) 21 cm emission observations of a third low redshift dla , the @xmath0 absorber towards the qso hs 1543 + 5921 .
the damped absorption has been identified as arising in a low surface brightness galaxy sbs 1543 + 593 ( @xcite ) .
hi emission has also been detected from this galaxy using the bonn telescope ( @xcite ) .
the extremely low redshift of the absorber allows us to make , for the first time , spatially resolved images of the 21 cm emission ; besides the hi mass , this also enables us to determine the velocity field of the galaxy and , hence , to estimate its dynamical mass .
the gmrt observations are detailed in sect .
[ sec : obs ] , while the results are presented in sect .
[ subsec : results ] and discussed in sect .
[ subsec : discuss ] . throughout the paper ,
we use a hubble constant of h@xmath20= 75 km s@xmath13 mpc@xmath13 , i.e. a distance to sbs 1543 + 593 of 38 mpc .
sbs 1543 + 593 was observed on 31 december 2000 using the gmrt .
the observing setup gave a total of 128 spectral channels over the total bandwidth of 2 mhz , corresponding to a channel separation of @xmath21 km s@xmath13 .
the standard calibrators 3c48 and 3c286 were observed at the start and end of the observing run and used to calibrate the visibility amplitudes and the bandpass shape .
phase calibration was done using observations of the nearby continuum sources 1438 + 621 and 1549 + 506 , one of which was observed once every 35 minutes .
the data were converted to fits from the raw telescope format and analysed in classic aips .
the brightest continuum source in the field of view is the steep spectrum source 7c 1543 + 5920 .
the flux as measured at gmrt is 42 mjy , in excellent agreement with that obtained in the nvss ( @xcite ) .
the source is at a distance of 2.2@xmath22 ( corresponding to an impact parameter of @xmath23 kpc ) from sbs 1543 + 593 .
no statistically significant 21 cm absorption is seen against this source ; the 3 @xmath17 limit to the optical depth after smoothing to a velocity resolution of 10 km s@xmath13 is @xmath14 . this yields column density limits of @xmath24 @xmath8 and @xmath25 @xmath8 for hi with spin temperatures of 100 k and 1000 k respectively , along this line of sight . as an aside
, we note that 7c 1543 + 5920 has a spectral index of @xmath26 and is , at best , marginally resolved on our longest baselines the upper limit on its angular size is @xmath27 .
given its relatively high galactic latitude ( @xmath28 ) , it is likely that 7c 1543 + 5920 is a compact steep spectrum ( css ) source .
the gmrt has a hybrid configuration ; 14 of its 30 antennas are located in a central compact array with size @xmath29 km ( @xmath30 k@xmath31 at 21 cm ) while the remaining antennas are distributed in the three arms of a y configuration , giving a maximum baseline of @xmath32 km ( @xmath33 k@xmath31 at 21 cm ) .
the baselines obtained from antennas in the central square are similar in length to those of the vla in its d configuration , while the arm antennas provide baselines similar in length to the b array of the vla .
a single observation with the gmrt hence yields information on both large and small angular scales .
the search for absorption against the compact source 7c 1543 + 5920 was made using all available baselines .
however , the hi emission from sbs 1543 + 593 is heavily resolved on the longest baselines . to study this emission ,
cubes were hence made at a variety of uv ranges , viz .
0 - 5 k@xmath31 , 0 - 10 k@xmath31 and 0 - 20 k@xmath31 , corresponding to spatial resolutions of 39@xmath34 , @xmath35 and @xmath36 .
the two low resolution cubes were cleaned , using the aips task imagr . at the highest resolution , the signal to noise ratio is too low and the emission too diffuse for clean to work reliably .
however , at the signal to noise ratio of the map at this resolution , convolution by the dirty beam does not greatly degrade the dynamic range ( or image fidelity ) .
the morphology should hence be accurately traced , apart from an uncertainty in the scaling factor ( which is discussed in more detail below ) .
[ fig : mom0 ] shows maps of the integrated hi emission ( obtained using the aips task momnt ) at the three different resolutions discussed above .
the qso position is marked by a cross in all three maps .
the centre of the optical galaxy is very close to the qso position ; the qso is located only @xmath37 nne of the galaxy centre ( @xcite ) .
the low resolution map shows that the hi distribution is asymmetric with the peak emission displaced to the south - east of the galaxy centre .
the higher resolution maps show clearly that the hi distribution is concentrated in a ring - like structure , with hi emission actually being depressed at the centre of the galaxy .
the hi ring is coincident with the faint spiral arms seen in the optical images ( @xcite ) .
many of the irregular patches of emission seen in the hst image ( @xcite ) also appear to be associated with peaks of hi emission .
the hi concentration at @xmath38 corresponds to the hii region whose spectrum is given in reimers & hagen ( 1998 ) .
the velocity we measure at the location of this hii region is @xmath39 km s@xmath13 .
in addition to the inner hi ring , there are also spurs in the hi emission ( see figs .
[ fig : mom0][b ] & [ c ] ) towards the north and south ; these may mark the beginning of faint outer spiral arms .
measurement of the integrated flux corresponding to weak extended line emission can be a non - trivial problem ( see , for example , @xcite ) .
however , as a rule of thumb , ( apart from serious zero spacing problems ) , deeply cleaned images give a fairly reliable estimate of the total flux .
the integrated fluxes that we get from the cleaned @xmath40 and @xmath41 data cubes are @xmath42 jy km s@xmath13 and @xmath43 jy km s@xmath13 .
these are in excellent agreement with the single dish measurement of @xmath42 jy km s@xmath13 ( @xcite ) . from the area of the clean beam in these two low resolution images , the conversion from 1
jy beam@xmath13 km s@xmath13 to atoms @xmath8 is @xmath44 atoms @xmath8 and @xmath45 atoms @xmath8 respectively . since the highest resolution map ( fig . [ fig : mom0][c ] )
is uncleaned , there is no obvious way to convert the units from jy beam@xmath13 to jy .
if one simply uses the area of the best fit gaussian to the main lobe of the dirty beam to convert from jy beam@xmath13 to jy , one gets a total flux of @xmath46 jy km s@xmath13 .
this must be an overestimate , since it is clear from fig .
[ fig : mom0][c ] that a considerable fraction of the smooth emission seen in fig .
[ fig : mom0][a ] has been resolved out but the estimated flux in the higher resolution image is _ higher _ than that in the lower resolution ones . given this problem in scaling for the highest resolution image
, we estimate the column density at the qso position only from the two lower resolution maps .
the column densities are @xmath47 atoms @xmath8 and @xmath48 atoms @xmath8 for the @xmath49 and the @xmath41 resolutions , respectively .
the integrated emission profile of sbs 1543 + 593 , as derived from the @xmath50 resolution data ( smoothed to a velocity resolution of 10 km s@xmath13 ) , is shown in fig .
[ fig : mom0][d ] .
the systemic velocity as measured from the profile is 2862@xmath51 10 km s@xmath13 .
this agrees ( within the errors of the two measurements ) with the velocity of
2868@xmath51 2 km s@xmath13measured from the bonn spectrum by @xcite .
the velocity field of sbs 1543 + 593 ( derived using the aips task momnt on the @xmath41 resolution cube ) is shown in fig .
[ fig : mom1][a ] .
the iso - velocity contours are asymmetric ; they are straight on the approaching side and curved on the receding side .
this type of velocity field has been dubbed ` kinematically lopsided ' ( @xcite ) . as noted earlier ,
the galaxy is also morphologically lopsided .
the rotation curve was derived separately for the approaching and receding sides using the aips task gal . during these fits ,
the galaxy centre was kept fixed at the optical centre .
the systemic velocity was kept fixed at the value of 2870 km s@xmath13 , the inclination at @xmath52 and the position angle ( of the receding half of the major axis , measured east of north ) at @xmath53 ( all of which were obtained from an initial global fit to the velocity field ) .
the inclination and the position angle are in good agreement with our estimates from the optical image presented in reimers & hagen ( 1998 ) .
the derived rotation curves are shown in fig .
[ fig : mom1][b ] .
as expected from the iso - velocity contours , the rotation curve is almost linear on the approaching side , while it tends to flatten out on the receding side .
the maximum ( inclination corrected ) rotation speed in sbs 1543 + 593 is @xmath54 km s@xmath13 .
the rotation curve can be measured out to a radius of @xmath55 , corresponding to a linear distance of 11 kpc .
the implied dynamical mass is then m@xmath56 m@xmath4 .
the flux that we measure for sbs 1543 + 593 corresponds to a total hi mass of @xmath57m@xmath4 , in excellent agreement with the single dish measurement of bowen et al .
( 2001b ) .
this is thus the third dla whose hi mass is significantly less than that of an l@xmath58 spiral . besides these
, the galaxy ngc 4203 ( which lies in front of the qso ton 1480 , and is likely to be a dla ; @xcite ) has also been found to have a low hi mass .
thus , all low redshift dlas ( or candidate dlas ) for which hi emission observations have been attempted have masses less than that of an l@xmath19 spiral .
the derived hi masses are also considerably less than the m@xmath59 ( @xmath60 m@xmath4 ) obtained from schecter function fits to the hi mass function determined in blind hi surveys ( @xcite ) .
it also follows from these mass functions that large galaxies make the major contribution to the local hi mass density .
however , in order to determine the contribution of galaxies of different hi masses to the cross - section for dla absorption , one also needs to know the typical sizes of their hi disks .
rao & turnshek ( 1998 ) deduced , based on determinations of the typical sizes of hi disks made by rao ( 1994 ) , that bright spiral galaxies make the major contribution to the @xmath11 cross - section for dla absorption .
however , more recent determinations of the hi size and mass distribution of galaxies indicate that the cross - section for dla absorption is , in fact , not dominated by large spiral galaxies ( @xcite ) , and that a sample of low redshift dlas should contain a large variety of galaxy types .
this is consistent with the optical and hi observations of low @xmath61 dlas .
it would be interesting to compare the velocity profiles of low ionization metal lines in the hs 1543 + 5921 spectrum with the large scale kinematics of sbs 1543 + 593 .
unfortunately , high velocity resolution optical / uv observations are as yet not available for this system .
we note , however , that the systemic velocity that we obtain for sbs 1543 + 593 is in excellent agreement with the value of @xmath62 km s@xmath13 obtained from the single dish hi profile ( @xcite ) .
further , the velocity obtained in our observations ( @xmath63 km s@xmath13 ) for the gas coincident with the hii region ( for which an optical emission spectrum exists ) is in reasonable agreement with that obtained in the optical ( @xmath18 2700 km s@xmath13 ; @xcite ) , given the poor resolution ( @xmath64 angstroms , i.e. @xmath65 km s@xmath13 ) of the optical spectrum .
similarly , the velocity we obtain at the qso location is @xmath66 km s@xmath13 , which agrees within the error bars with the velocity of @xmath67 km s@xmath13 obtained from the ly@xmath68 line ( @xcite ) , given the large uncertainty ( @xmath69 km s@xmath13 ) in the latter measurement ( @xcite ) .
a high resolution low ionization metal line absorption spectrum of the qso ( and of ton 1480 , which lies behind ngc 4203 ) would provide interesting spot checks of the reliability in using such absorption lines to probe large scale gas kinematics of the absorbing galaxies ( see , for example , @xcite ) .
it is also interesting to compare the hi column density as obtained from the ly-@xmath1 spectrum with that obtained from the 21 cm emission .
the comparison is , however , complicated by the comparatively large size of the gmrt synthesized beam . at a distance of 38 mpc , @xmath70 corresponds to a linear size of @xmath71 kpc .
given this great disparity in the transverse sizes covered by the hi and uv measurements , the column density of @xmath72 atoms @xmath8 obtained from the 21 cm map is in reasonable agreement with the estimate of @xmath73 atoms @xmath8 obtained from the lyman-@xmath1 profile .
comparisons of dla samples with samples of galaxies selected by blind hi emission are complicated by issues of optical extinction .
it has long been suggested that dla samples could be seriously biased against high hi column density systems , because such absorbers would contain enough dust to substantially dim the background qso ( e.g. @xcite ) .
deep optical observations of a radio selected sample of qsos however suggest that such biases may be modest ( @xcite ) .
it is curious ( but , of course , not statistically significant , particularly given the fact that optical depth effects may be less important in lsb galaxies than in normal spirals ) that the qso hs 1543 + 5921 lies behind a region of low local hi column density in sbs 1543 + 593 .
interestingly , an inspection of the hi maps of the s0 galaxy ngc 4203 presented by van driel et al .
( 1988 ) shows that ton 1480 also lies behind a region of low column density .
unfortunately , testing the above bias would require 21 cm mapping of a statistically significant number of dlas which is , alas , a task for the next generation radio telescope .
the gmrt observations presented in this paper would not have been possible without the many years of dedicated effort put in by the gmrt staff to build the telescope .
the gmrt is run by the national centre for radio astrophysics of the tata institute of fundamental research .
useful discussions with r. nityananda are gratefully acknowledged .
we also thank the referee , e. brinks , for several comments which improved both the readability and the content of this paper .
bowen d. v. , tripp t. m. , jenkins e. b. , 2001a , aj , 121 , 1456 bowen d. v. , huchtmeier w. , brinks e. , tripp t. m. , jenkins e. b. , 2001b , a&a , 372 , 820 centurin m. , bonifacio p. , molaro p. , vladilo g. , 2000 , apj 536 , 540 chengalur j. n. , kanekar n. , 2000 , mnras 318 , 303 condon j. j. , cotton w. d. , greisen e. w. , et al . , 1998 , aj , 115 , 1693 .
ellison s. l. , yan l. , hook i. m. , et al .
, 2001 , to appear in a&a ( astro - ph/0109205 ) fall s. m. , pei y. c. , 1993 , apj , 402 , 479 haehnelt m. g. , steinmetz m. , rauch m. , 1998 , apj , 495 , 64 .
heisler j. , ostriker j. p. , 1988 ,
apj , 332 , 543 jrster s. , van moorsel g. a. , 1995 , aj , 110 , 2037 kanekar n. , chengalur j. n. , 2001 , a&a 369 , 42 kanekar n. , chengalur j. n. , subrahmanyan r. , petitjean p. , 2001
, a&a , 367 , 46 lane , w. , 2000 , ph.d .
thesis , univ . of groningen le brun v. , bergeron j. , boiss p. , deharveng j .-
, 1997 , a&a , 321 , 733 .
mcdonald p. , miralda - escude j. , 1999 , apj , 519 , 486 .
miller e. d. , knezek p. m. , bregman j. n. , 1999 , apj , 510 , l95 nestor d. b. , rao s. m. , turnshek d. a. , et al .
, 2001 , to appear in `` extragalactic gas at low redshift '' ( astro - ph/0108146 ) pei y. c. , fall s. m. , hauser m. g. , 1999 , apj , 522 , 604 pettini m. , ellison , s. l. , steidel c. c. , bowen d. v. , 1999 , apj 510 , 576 prochaska j. x. , wolfe a. m. , 1997 , apj , 487 , 73 prochaska j. x. , wolfe a. m. , 1998 , apj , 507 , 113 rao s. m. , 1994 , ph.d .
thesis , university of pittsburgh rao s. m. , turnshek , d. a. , 1998 , apj , 500 , l115 .
reimers d. , hagen h .- j .
, 1998 , a&a , 329 , l25 rupen , m. p. , 1999
, asp conf .
ser . 180 : synthesis imaging in radio astronomy ii , 229 rosenberg j. l , schneider s. e. , 2001 , to appear in `` extragalactic gas at low redshift '' ( astro - ph/0107495 ) storrie - lombardi l. j. , mcmahon r. g. , irwin m. j. , 1996 , mnras , 283 , l79 . storrie - lombardi l. , wolfe a. m. , 2000 , apj 543 , 552 swaters r. a. , schoenmakers r. h. m. , sancisi r. , van albada t. s. , 1999 , mnras , 304 , 330 van driel w. , van woerden h. , gallagher j. s. iii , schwarz u. j. , 1988 , a&a , 191 , 201 wolfe a. m. , turnshek d. a. , smith h. e. , cohen r. d. , 1986 , apjs , 61 , 249 .
zwaan m. , briggs f. h. , verheijen m. , to appear in `` extragalactic gas at low redshift '' ( astro - ph/0108498 ) zwaan m. a. , briggs f. h. , sprayberry d. sorar e. 1997 , apj 490 , 173 | we present giant metrewave radio telescope ( gmrt ) hi 21 cm emission images of the @xmath0 damped lyman-@xmath1 ( dla ) absorber towards the qso hs 1543 + 5921 .
the dla has been earlier identified as a low surface brightness galaxy sbs 1543 + 593 ( @xcite ) with small ( @xmath2 pc ) impact parameter to the qso line of sight .
the extremely low redshift of the absorber allows us to make , for the first time , spatially resolved images of the 21 cm emission ; besides the hi mass , this also enables us to determine the velocity field of the galaxy and , hence , to estimate its dynamical mass .
we obtain a total hi mass of @xmath3 m@xmath4 , considerably smaller than the value of m@xmath5 determined from blind 21 cm emission surveys .
this continues the trend of low hi mass in all low redshift dlas for which hi emission observations have been attempted .
we also find that the qso lies behind a region of low local hi column density in the foreground galaxy .
this is interesting in view of suggestions that dla samples are biased against high hi column density systems .
finally , the dynamical mass of the galaxy is found to be m@xmath6 m@xmath4 . |
it is classically well known that a complete toric variety of dimension @xmath3 is projective ( * ? ? ?
* , prop.8.1 ) . for higher dimension
this fact is not true , as shown by lot of counterexamples , the first of which was given by m.demazure @xcite . for smooth complete toric varieties , p.kleinschmidt and b.sturmfels
@xcite proved that for picard number ( in the following also called the _ rank _ ) @xmath4 they are projective in every dimension .
this result can not be extended to higher values of the rank , as shown by a famous example given in oda s book @xcite : this is a smooth complete 3dimensional toric variety @xmath5 of rank @xmath6 , such that @xmath7 inside @xmath8 : therefore @xmath5 admits non trivial numerically effective classes ( among which the anti canonical one ) but does not admit any ample class .
when dropping the smoothness hypothesis , kleinschmidt sturmfels bound does no more hold even for @xmath0factorial singularities : a counterexample has been given by f.berchtold and j.hausen ( * ? ? ?
* ex.10.2 ) who produced a @xmath0factorial complete 3dimensional toric variety @xmath5 of rank @xmath9 such that @xmath10 is a 1dimensional cone generated by the anti
canonical class .
this example is essentially produced by suppressing a fan generator in the oda s example ( see remark [ rem : ] ) .
nevertheless , a @xmath0factorial complete toric variety of rank @xmath11 is always projective as a quotient of a weighed projective space ( so called a _ fake _ wps ) @xcite , @xcite .
it is then natural to ask what happens for @xmath12 : surprisingly , the current literature does not seem to give any answer to this question , at least as far as we know .
the first and main result we are going to present in this paper is a extension of the kleinschmidt
sturmfels result to @xmath0factorial setup , by decreasing the bound on the rank @xmath13 , namely * theorem [ teo : erredueproj ] * _ every @xmath0factorial complete toric variety of picard number @xmath14 is projective . _ our proof is essentially obtained by generalizing a berchtold
hausen s argument proving the last part of ( * ? ? ?
* prop.10.1 ) , by means of @xmath1linear gale duality , as developed in and @xcite .
in fact @xmath1linear gale duality allows us to observe the general fact , summarized by lemma [ lem : dueparte ] , about the mutual position of pairs of cones gale dually associated with pairs of maximal cones of the fan admitting a common facet .
when the rank is 2 , lemma [ lem : dueparte ] gives rise to strong consequences proving theorem [ teo : erredueproj ] .
this proof turns out to be very easy and we believe this is a further reason of interest .
a second result of the present paper is proposed in the last section [ sez : esempio ] , by exhibiting an example of @xmath0factorial complete and non projective 3dimensional toric variety @xmath5 with rank @xmath9 and _ not admitting any non
trivial numerically effective divisor _
i.e. such that @xmath15 .
more precisely , theorem [ teo : erredueproj ] implies that @xmath16 example given in [ sez : esempio ] then shows that condition ( [ nef0q ] ) is sharp .
actually this is not a new example .
in fact o.fujino and s.payne proved that @xmath17 , for a smooth and complete toric 3fold @xmath5 @xcite : their example 1 is obtained by blowing up a complete toric variety which turns out to be isomorphic to the one given in
[ sez : esempio ] .
this example jointly with theorem [ teo : erredueproj ] allows us to conclude the sharpness of condition ( [ nef0q ] ) , which is a new result .
let us finally notice that this example can be produced by _ breaking the symmetry _ of the above mentioned berchtold
hausen example , just _ deforming _
one generator of the effective cone @xmath18 inside @xmath19 . for more details
see remark [ rem : ] .
we would like to thank cinzia casagrande for helpful conversations and suggestions .
we also thank daniela caldini for her contribution in making the figures of the present paper .
in the present paper we deal with @xmath0factorial complete toric varieties associated with simplicial and complete fans . for preliminaries and used notation on toric varieties we refer the reader to and @xcite
. we will also apply @xmath1linear gale duality as developed in .
every time the needed nomenclature will be recalled either directly by giving the necessary definition or by reporting the precise reference . here is a list of main notation and relative references : + let @xmath20 be a @xmath21dimensional toric variety and @xmath22 the acting torus , then @xmath23 let @xmath24 be a @xmath25 integer matrix , then @xmath26 given a matrix @xmath27 then @xmath28
let @xmath20 be a @xmath0factorial complete toric variety of dimension @xmath21 and picard number @xmath13 ( in the following also called _ the rank of @xmath5 _ ) .
then @xmath29 is a rational , simplicial and complete fan in @xmath30 .
its 1skeleton is given by @xmath31 , where @xmath32 is a generator of the monoid associated with the corresponding ray .
the @xmath33 integer matrix @xmath34 will be called _ a fan matrix _ of @xmath5 and it turns out to be a reduced @xmath35matrix , in the sense of and ( * ? ? ?
* def . 1.2 ) . a maximal cone @xmath36 can be identified with a maximal rank @xmath37 submatrix @xmath38 of the fan matrix @xmath39 , assigned by all the columns of @xmath39 generating some ray in @xmath40 : in particular @xmath41 denotes the position of those columns inside @xmath39 , while @xmath42 is the complementary subset .
we will also write @xmath43 .
let @xmath44 be a maximal rank @xmath45 _ gale dual _ matrix of @xmath39 i.e. such that @xmath46 and define the following set of _ dual _ cones inside @xmath47 : @xmath48 @xmath49 turns out to be a _ bunch of cones _ in the sense of @xcite ( see also @xcite ) .
in particular [ thm : camera ] ( * ? ? ?
* prop.15.2.1(c ) ) in the above notation , the khler cone of the @xmath0factorial complete toric variety @xmath20 is given by @xmath50 in particular @xmath5 is projective if and only @xmath51 .
let us set @xmath52 then we get the following : [ lem : dueparte ] with the notation introduced above : * if @xmath53 then for every @xmath54 there exists a unique @xmath55 such that @xmath56 * let @xmath57 be as in part a ) : then the vectors @xmath58 and @xmath59 lie on the same side with respect to the hyperplane @xmath60 generated by @xmath61 .
\a ) this is the dual assertion of the fact that every facet of @xmath62 belongs to exactly two maximal cones of @xmath62 . since @xmath63 , @xmath64 .
then the facet @xmath65 of @xmath66 opposite to @xmath67 belongs to exactly one other maximal cone @xmath68 .
if @xmath69 is the vector opposite to @xmath65 in @xmath68 then ( [ eq : iprime ] ) holds .
+ b ) let @xmath70 be the hyperplane supporting the common face @xmath71 : it is generated by vectors @xmath72 . since @xmath67 and @xmath73
lie on opposite sides of @xmath74 , the cone generated by @xmath67 and @xmath73 intersects @xmath74 giving a relation @xmath75 such that @xmath76 .
therefore in @xmath77 there is a vector having positive entries at places @xmath78 and @xmath79 and @xmath80 at places in @xmath81 .
it corresponds to a vector @xmath82 which is clearly a normal vector to the hyperplane @xmath83 . to better understand the argument proving lemma [ lem : dueparte ] it may be of some help considering the easy example given by @xmath84 .
a fan matrix @xmath39 of @xmath5 and a gale dual @xmath85 of @xmath39 are the following @xmath86 in this case @xmath87 and part a ) is clear .
let us now consider the two maximal cones @xmath88 and @xmath89 , whose common facet is given by @xmath90 .
in the notation of lemma [ lem : dueparte ] one has @xmath91 : the argument proving part
b ) gives @xmath92 and relation ( [ relazione ] ) is given by @xmath93 the associated vector in @xmath77 turns out to be the first row of @xmath85 , hence giving @xmath94 lying on the same side with respect to the hyperplane ( i.e. the line ) @xmath95 spanned by the fourth column @xmath96 of @xmath85 .
cones of the bunch @xmath49 as in lemma [ lem : pqr].b ) ]
given a simplicial and complete fan @xmath29 , if @xmath12 then @xmath97 for every @xmath98 .
the following is a direct application of lemma [ lem : dueparte ] .
[ lem : pqr ] let @xmath99 be three distinct vectors giving as many columns of @xmath85 , such that @xmath100 .
* exactly one between @xmath101 and @xmath102 is in @xmath49 . * if @xmath103 then @xmath104 and @xmath105 .
\a ) this is a direct consequence of lemma [ lem : dueparte].a ) by setting @xmath106 and @xmath107 .
\b ) if @xmath108 then @xmath109 ( see fig . [ fig1 ] ) . by
a ) exactly one between @xmath101 and @xmath102 belongs to @xmath49 and either @xmath110 or @xmath111 . by lemma [ lem : dueparte].b )
the former can not occur , meaning that @xmath112 .
we are now in a position to prove the main result of the present paper .
[ teo : erredueproj ] every @xmath0-factorial complete toric variety with picard number @xmath12 is projective .
consider a cone @xmath113 .
if there exists a further column @xmath114 of @xmath85 such that @xmath115 then lemma [ lem : pqr].a ) ensures that either @xmath116 or @xmath117 belongs to @xmath49 .
then , by repeatedly applying lemma [ lem : pqr].a ) , we get a _ minimal
_ cone @xmath118 , in the sense that there are no columns of @xmath85 in @xmath119 .
we can then assume @xmath120 is a minimal chamber : we claim that @xmath121 is contained in every cone @xmath122 .
in fact , if one of @xmath123 is a ray of @xmath124 then , up to renaming @xmath125 and @xmath126 , we can assume @xmath127 .
lemma [ lem : dueparte].b ) guarantees that either @xmath128 and we are done or @xmath129 , against the minimality of @xmath130 .
then we can assume @xmath131 , with @xmath132 .
assume @xmath133 .
by the minimality of @xmath121 , the cone generated by @xmath134 has a ray in @xmath135 and a ray in @xmath136 .
we may then assume @xmath137 , as in fig .
consider the three vectors @xmath138 : since @xmath139 and @xmath140 , lemma [ lem : pqr].b ) ensures that @xmath141 .
finally consider the three vectors @xmath142 : since both @xmath143 and @xmath144 belong to @xmath49 we get a contradiction with lemma [ lem : dueparte].b ) .
then , recalling theorem [ thm : camera ] , @xmath145 is a 2dimensional cone , meaning that @xmath5 is projective .
cones of the bunch @xmath49 as in theorem [ teo : erredueproj ] ]
in this section we are going to exhibit an example of a 3dimensional @xmath0factorial complete toric variety @xmath5 of picard number @xmath9 not admitting _ any non
trivial numerically effective divisor _
i.e. such that @xmath15
. therefore such an example is significant for at least two reasons : * it gives a sharp counterexample to the question if theorem [ teo : erredueproj ] is holding for higher values of @xmath13 : in this sense it does not say nothing new with respect to the berchtold
hausen example ( * ? ? ?
* ex.10.2 ) ; * for smooth complete toric @xmath146folds o.fujino and s.payne @xcite proved that if @xmath15 then @xmath147 : their example 1 is produced by performing a double divisorial blow up of a complete toric variety which turns out to be isomorphic to the one we are going to present , showing that this last inequality does no more hold when the smoothness hypothesis is replaced by the @xmath0factorial one , actually giving the sharpness of condition ( [ nef0q ] ) in the introduction .
let us start by considering the following positive @xmath148 @xmath149matrix , ( * ? ? ?
1.4 ) @xmath150 the cone @xmath151 spanned by its columns gives the effective cone @xmath18 of a @xmath0factorial complete 3dimensional toric variety @xmath5 of picard number @xmath9 whose fan matrix is given by a gale dual @xmath35matrix @xmath39 of @xmath85 , ( * ? ? ?
1.2 ) e.g. @xmath152 this fan matrix can support 8 different rational , simplicial and complete fans whose 1skeletons coincide with the rays spanned by all the columns of @xmath39 , namely @xmath153 with @xmath154 where @xmath157 denotes the cone spanned in @xmath30 by @xmath158th , @xmath78th and @xmath79the columns of the fan @xmath39 . fans @xmath159 , for @xmath160 , give rise to projective toric varieties ,
each of them being associated with a full dimensional chamber inside the moving cone @xmath161 of @xmath5 , which actually depends on the 1skeleton of the fan , only , i.e. on the columns of @xmath39 : the situation is represented in fig .
[ fig3 ] , by cutting out the positive orthant @xmath162 , inside @xmath163 , with the hyperplane @xmath156 . on the contrary
the last fan @xmath164 does not give rise to a projective toric variety since its gale dual bunch of cones is given by @xmath165 where now @xmath157 denotes the cone spanned in @xmath166 by @xmath158th , @xmath78th and @xmath79the columns of the weight matrix @xmath85 .
one can easily check that the intersection of all the cones in the bunch @xmath167 is trivial , hence giving @xmath168 for sake of completeness , let us finally observe that , using technics like those given in ( * ? ? ?
* thm . 4 )
, one can check that the anti - canonical class lies on the wall separating chamber 2 and 7 in fig .
[ fig3 ] : in particular none of the fans in @xmath169 give a gorenstien toric varieties but @xmath170 and @xmath171 are _ weak @xmath0-fano _ toric varieties .
[ rem : ] the previous example has been obtained by _ deforming _ the berchtold
hausen example ( * ? ? ?
* ex.10.2 ) in the following sense .
the latter can be obtained as above , by starting from the weight matrix @xmath172 also in this case we obtain 8 rational , simplicial and complete fans whose fan matrix is @xmath174 , given by the @xmath175 possible subdivision of the three quadrangular faces of the prism whose vertexes are given by the columns of @xmath174 ( see fig.[fig5 ] ) .
but only 6 of them give rise to projective varieties , while the remaining 2 fans originates two distinct complete and non projective @xmath0factorial 3dimensional varieties of picard number 3 .
the situation is represented in fig.[fig5 ] and gale dually in fig .
explicitly the two non projective fans are the following : @xmath176 whose gale dual bunches are given by @xmath177 looking at fig .
[ fig4 ] , one can easily check that @xmath178 giving the 1dimensional cone generated by the class @xmath179 of the anti
canonical divisor .
[ fig4 ] explains as the anti - canonical class lies on the intersection of all the chambers , so giving that every fan in @xmath180 gives a non - gorestein weak @xmath0-fano toric variety . the two symmetric subdivisions of the prism , whose vertexes are given by the columns of @xmath39 , and giving rise to two distinct and bi
meromorphic @xmath0factorial complete and non projective toric varieties @xmath20 and @xmath181 , respectively .
the former is that given by ( * ? ? ?
* ex.10.2 ) .
notice that all the possible fans in @xmath169 are given by the @xmath175 possible subdivisions of the quadrangular faces .
] the two fans @xmath29 and @xmath182 corresponds to symmetric subdivisions of the three quadrangular faces of the prism generated by the columns of @xmath39 ( see fig.[fig5 ] ) .
such a symmetry is gale dually reflected by the position of the anti
canonical class with respect the subdivision of @xmath183 giving the secondary fan .
more precisely , fig.[fig4 ] clearly shows that @xmath183 and the secondary fan turn out to be invariant under the action of the three symmetries with respect to the three axis of the triangle representing the section of @xmath155 with the plane @xmath156 , i.e. the symmetries of @xmath163 with respect to the three planes supporting the 2dimensional cones @xmath184 , @xmath185 and @xmath186 .
these symmetries correspond to permutations on the columns of the weight matrix @xmath85 given by @xmath187 , @xmath188 and @xmath189 , respectively .
notice that bunches @xmath190 and @xmath191 are related each other by every one of those permutations .
geometrically this means that @xmath5 and @xmath192 are related by a bi
meromorphic map , still called a _ flip _ , given by the contraction of three facets of the fan , namely the closures of torus orbits @xmath193 , @xmath194 and @xmath195 , followed by the small resolution whose exceptional locus is given by the union of the symmetric facets @xmath196 , @xmath197 and @xmath198 ( compare with fig.[fig5 ] ) .
then @xmath5 and @xmath192 are isomorphic in codimension 1 , but no in codimension 2 . if we now consider the weight matrix @xmath199 with @xmath200 a positive integer number , we get the berchtold hausen example for @xmath201 , our example for @xmath202 and an infinite number of distinct examples analogous to the latter , for @xmath203 .
if we look at the evolution of the fan @xmath204 , hence of the chamber 7 in fig .
[ fig3 ] , we get a countable family @xmath205 of @xmath0factorial projective toric varieties degenerating to a non projective complete one for @xmath201 .
let us finally observe that a similar situation , although with picard number 4 , can be observed for the oda s example @xcite too , from which the berchtold
hausen one is obtained by suppressing a fan generator . in oda s
notation , the symmetric non projective fan , with respect to the one presented in @xcite , can be obtained by exchanging @xmath206 and @xmath207 . ,
15 of _ ergebnisse der mathematik und ihrer grenzgebiete ( 3 ) [ results in mathematics and related areas ( 3)]_. springer - verlag , berlin , 1988 .
an introduction to the theory of toric varieties , translated from the japanese . | this paper is devoted to settle two still open problems , connected with the existence of ample and nef divisors on a @xmath0factorial complete toric variety .
the first problem is about the existence of ample divisors when the picard number is 2 : we give a positive answer to this question , by studying the secondary fan by means of @xmath1linear gale duality .
the second problem is about the minimum value of the picard number allowing the vanishing of the @xmath2 cone : we present a 3dimensional example showing that this value can not be greater then 3 , which , under the previous result , is also the minimum value guaranteeing the existence of non projective examples . |
in our current working picture of active galactic nuclei ( agns ) the underlying power source is thought to be a supermassive black hole , which is accreting matter from the host galaxy .
the accretion flow very close to the central object is believed to form an equatorial accretion disk , by analogy with stellar accretion - powered systems such as cataclysmic variables and low - mass x - ray binaries .
the presence of accretion disks in agns , although appealing from a theoretical perspective and generally assumed , had received only limited and indirect observational support until a few years ago , when direct dynamical evidence became available .
this evidence comprises double - peaked line profiles characteristic of matter rotating in a disk , much like the double - peaked emission lines of cataclysmic variables ( see , for example , young & schneider 1980 ; young , schneider , & shectman 1981 ; marsh 1988 ) .
a number of authors ( e.g. , oke 1987 ; prez et al . 1988 ; chen , halpern & filippenko 1989 ; chen & halpern 1989 ; halpern 1990 ) had proposed that the double - peaked balmer lines found in the optical spectra of a handful broad - line radio galaxies such as 3c 390.3 , arp 102b , and 3c 332 originate in accretion disks around supermassive black holes at the centers of these galaxies .
we thus carried out a spectroscopic survey of almost 100 moderate - redshift broad - line radio galaxies and radio loud quasars ( eracleous & halpern 1994 , hereafter paper i ) and discovered 19 new double - peaked balmer lines of which a dozen conformed with a simple , kinematic accretion disk model .
we also found that such objects stand out on the basis of additional spectroscopic properties of their hosts .
independent dynamical evidence for a disk - like accretion flow in the immediate vicinity of the central black hole ( within a hundred gravitational radii ) has been provided by x - ray spectroscopy of seyfert galaxies with .
the profiles of the fe k lines of almost all seyfert galaxies observed with are extremely broad and asymmetric with full widths at zero intensity approaching a third of the speed of light ( mushotzky et al .
1995 ; tanaka et al . 1995 ; nandra et al .
1997 ) . in most cases the line profiles can be described very well by models attributing the emission to the inner parts of a relativistic accretion disk .
we have continued to observe radio - loud agns is search of double - peaked emitters and in order to extend our tests of models for the origin of their broad balmer lines . in this paper
we report the completion of the survey begun in paper i and the refined conclusions from it .
we make use of new spectroscopic data that we have obtained as well as a great deal of information on double - peaked emitters from recent multi - wavelength surveys .
we also take into consideration some of the variability properties of double - peaked emitters , especially the results of reverberation of 3c 390.3 . because emission lines from agn accretion disks are
a tool for studying the dynamical and thermal behavior of the disk , one of our long - term goals is to pursue such studies by monitoring the variability of double - peaked emitters that we have found in this survey .
another , more lofty , goal is to use the double - peaked emission lines ( and their variability ) to test dynamical models for the line - emitting gas in agns , including , for example , bipolar radial flows , and the binary black hole hypothesis .
the confirmation or rejection of _ any _ model for agn broad - line regions would represent progress since the dynamics of the line - emitting gas are still poorly understood . to this end
we study the general properties of double - peaked emitters and compare them with those of the average radio - loud agn .
we use these properties in combination with the profiles and relative strengths of the emission lines to assess the applicability of various proposed scenarios for the origin of the double - peaked lines .
we find that the accretion - disk interpretation of the line profiles to be the most appealing because it provides a self - consistent framework within which _ all _ of the properties of the hosts can be understood . throughout this paper
we adopt the following nomenclature : we refer to emission lines with double peaks or twin shoulders as double - peaked lines and to the objects that display them as double - peaked emitters . on occasion
, we distinguish the subset of double - peaked lines that can be fitted well with a model of a relativistic , keplerian disk ( see 3 ) by referring to them as disk - like lines ; their hosts are referred to as disk - like emitters .
we adopt a hubble constant of @xmath0 and a deceleration parameter of @xmath1 .
the target selection strategy and the motivation behind it are described in detail paper i. in summary , we chose to observe all moderate - redshift ( @xmath2 ) , radio - loud agns for which good hspectra were not available in the literature . the collection of targets was supplemented with objects ( listed in paper i ) whose spectra were drawn from the literature . in paper
i we reported the observation of 74 broad - lined objects from the original target list , while here we present spectra of another 10 certified broad - lined objects .
we also report supplementary observations of 6 objects whose h spectra were presented in paper i. these observations were carried out in order to cover a broader spectral range including their broad h lines and , if possible , also their narrow [ ] 3727 and broad 2800 lines .
finally , we also present spectra of 3 objects with double - peaked emission lines , which had been originally observed by other authors but had not been studied in detail ( cbs 74 , gonalves , vron , & vron - cetty 1998 ; pks 0921213 , simpson 1994 ; cso 643 , maxfield , djorgovski , & thompson 1995 ) .
thus , the final collection of radio - loud , broad - lined agns consists of 85 objects observed by us plus 19 objects whose spectra were drawn from the literature ( the spectrum of pictor a used in paper i was taken from the literature , while in this paper we include a spectrum obtained by us and presented in halpern & eracleous 1994 ) . in the process of acquiring the data we also observed 12 objects which had only narrow emission lines and seven objects whose redshifts were higher than 0.4 ( see discussion in paper i , and below ) .
the final set of object is not a complete sample in a statistical sense , but rather a representative collection of suitable objects in agn catalogs , _
circa _ 1991 , when the target selection was made .
the observations presented here were carried out over several observing runs using four different telescopes and spectrographs , namely the 4 m telescope and ritchie - chretien ( rc ) spectrograph and the 2.1 m telescope and goldcam spectrograph at kitt peak national observatory , the 4 m telescope and rc spectrograph at cerro tololo interamerican observatory , the 3 m shane telescope and kast double spectrograph at lick observatory , and the 2.4 m telescope at mdm observatory .
most observations were carried out between 1993 and 1995 and a small number of them were carried out between 1997 and 2000 .
the journal of observations is given in table [ tjour ] .
the spectra were taken through a narrow slit ( 1720 ) in moderate seeing conditions ( 1725 ) .
the slit was oriented along the parallactic angle whenever necessary to avoid differential loss of light resulting from atmospheric refraction .
the spectra were extracted from windows of typical width 48 along the slit and were calibrated in a standard fashion as described in paper i. the final spectral resolution was approximately 6 for the spectra taken with the 4 m telescopes and 3.54.5 for the spectra taken with the other telescopes .
our collection of newly - observed objects includes 13 agns with broad balmer lines ; their h spectra are shown in figure [ fspec ] .
the remaining 3 newly - observed objects ( pks 051148 , 3c 381 , and 3c 456 ) were found to have only narrow emission lines . finally , two objects ( pks 135512 , and pks 2312319 ) were found to have grossly incorrect cataloged redshifts ; as a consequence , their h lines did not fall within the observed spectral range .
the spectra of the narrow - line radio galaxies and the objects with incorrect redshifts are included in a companion paper in the _ astrophysical journal supplement series_. that paper also includes a table of accurate redshifts of all of the objects that we have observed in our survey .
as in paper i , we identify h profiles with displaced peaks or shoulders and divide them into groups accordingly .
the noteworthy objects from this paper are 4c 31.06 , pictor a , cbs 74 , pks 0921213 , and cso 643 .
our final sample includes 106 radio - loud objects of which 2023% have double - peaked balmer lines ( depending on whether or not we count objects whose double - peaked lines were known _ a priori _ ) . out of a total of 24 double - peaked emitters ,
17 have the blue peak stronger than the red , which has a chance probability of 0.02 .
thus , there is an intrinsic preference for the blue peak being stronger than the red , which disfavors scenarios in which a random distribution in phase space of line - emitting `` clouds '' gives rise to double - peaked line profiles by chance .
furthermore , 13 out of the 17 profiles that have the blue peak stronger than the red can be fitted well with a simple disk model , as we describe in later sections .
it is important to note that membership in the above groups of line profiles may be temporary , since double - peaked line profiles are known to vary significantly on time scales of years .
the variations can take the form of small but significant changes in the relative strengths of the two peaks ( e.g. , arp 102b ; miller & peterson 1990 ; newman et al .
1997 ) , complete reversals in the symmetry of the profile , namely the blue peak being stronger than the red at some epochs but not at others ( e.g. , 3c 390.3 and 3c 332 ; zheng , veilleux , & grandi 1991 ; gilbert et al .
1999 ) , and the emergence of double - peaked profiles with subsequent dramatic variations ( e.g. , pictor a ; halpern & eracleous 1994 ; sulentic et al . 1995a
; eracleous & halpern 1998 ) .
therefore , the relative numbers of blue- and red - asymmetric double - peaked profiles could reflect the fraction of time that the profiles spend in a given state .
for the purposes of this paper , we consider all double - peaked emitters as a group , without making a distinction based on the sense of their asymmetry or whether or not the simplest models can fit their profiles in detail . to study the broad emission - line profiles quantitatively , we measured the line widths and shifts at half maximum and at zero intensity .
the results of these measurements , are listed in table [ tstar ] .
the measured widths and shifts at zero intensity should be regarded with caution ; their error bars ( given in table [ tstar ] ) are an order of magnitude larger than those on the widths and shifts at half maximum because of the uncertainty in determining the continuum level and hence the uncertainty in identifying the far wings of the line . in figure [ fwidths ]
we compare the distributions of fwhm and fwzi of double - peaked emitters with those of other objects in our entire collection , following paper i. the mean values and the probabilities that corresponding distributions are drawn from the same parent population ( according to the kolmogorov - smirnov , or k - s , test ) are given in table [ tcomp ] .
the most significant difference is in the distributions of fwhm with double - peaked emitters having h lines that are on average twice as broad as other objects .
the distribution of fractional shifts among all objects in the collection is plotted in figure [ fshifts ] , in which the double - peaked emitters are represented by the shaded histogram bins .
there is a preference for redshifts , with a mean shift value of @xmath3 , or @xmath4 ( the same at half maximum and at zero intensity , both for double - peaked emitters and for other radio - loud agns ) . in 4/5 of double - peaked emitters
the h lines show a net redshift at half maximum . at zero intensity
the error bars are quite large , leading us to use extremely coarse bins . within these large error bars
there are no blueshifts among double - peaked emitters . our findings regarding the properties of the line profiles of the entire set of radio - loud agns are by no means new results .
the widths and shifts we measure are in agreement with the results of more systematic and detailed studies such as those of boroson & green ( 1992 ) , brotherton ( 1996 ) , corbin ( 1994 , 1997b ) , sulentic et al .
( 1995b ) , and marziani et al .
it is also noteworthy that variants of many of these results have been known for more than 20 years ( see miley & miller 1979 and steiner 1981 ) .
our comparison of the properties of double - peaked emitters and other objects shows that double - peaked emitters are extreme objects , occupying the high end of the balmer line width and redshift distributions .
we will return to these results in 6 where we will discuss them in the context of scenarios of the structure of line - emitting region .
we have tried to fit h profiles with well - defined twin peaks or pronounced shoulders ( 4c 31.06 , pictor a , cbs 74 , pks 0921213 , and cso 643 ) with the relativistic keplerian disk model of chen & halpern ( 1989 ) .
this model assumes that the line originates on the surface of a circular , keplerian disk whose axis is inclined by an angle @xmath5 relative to the line of sight , between radii @xmath6 and @xmath7 ( expressed in units of the gravitational radius , @xmath8 , where is the mass of the black hole ) .
the disk has an axisymmetric emissivity of the form @xmath9 .
local broadening of the line is represented by a gaussian rest - frame profile of velocity dispersion is @xmath10 ( this is a combination of turbulent motions in the disk and the velocity gradient within the cells used in our numerical integration )
. the free parameters of the model are thus @xmath11 and @xmath10 .
the flux illuminating the outer disk is expected to vary as @xmath12 , while the the h flux emerging from the disk is proportional to the illuminating flux for an extremely wide range of values of the latter ( see collin - souffrin & dumont 1989 and dumont & collin - souffrin 1990a , b ) .
therefore , we adopt @xmath13 as a reasonable prescription for the h emissivity of the disk .
lower values of @xmath14 can result when the h emissivity saturates at very high values of the ionizing flux . in figure [ ffit]a ,
b , e we show the h profiles of 4c 31.06 , cbs 74 , and pks 0921213 with the best - fitting disk models superposed .
the model parameters are generally in the same range as the those found from fitting the double - peaked profiles of paper i ; they are summarized in table [ tfit ] .
the goodness of the fit was judged by eye and the error bars on the model parameters were obtained by perturbing each parameter about its optimal value and adjusting all other parameters until an acceptable fit was no longer possible .
the above model does not describe the h profiles of pictor a and cso 643 as well as it does those of other objects . in the case of pictor
a , the red peak is stronger than the blue , contrary to the expectations of the simplest , axisymmetric disk models , while in the case of cso 643 an _ ad hoc _ blueshift of 1370 appears to be needed for a good fit .
therefore , we attempted to fit them with the elliptical disk model of eracleous et al .
( 1995 ) , which has two additional free parameters .
we are not suggesting that these objects necessarily harbor eccentric accretion disks , although we do consider this a plausible scenario .
the point of this exercise is to demonstrate that with some refinement , accretion disk models can account for a much wider variety of line profiles .
the model parameters for these two objects ( for both circular and elliptical disk models ) are included in table [ tfit ] and the fits are compared to the data in figure [ ffit]c , d , f , g .
we discuss the physical justification for these models further in [ s_indiv ] .
an important prediction of the models of photoionized accretion disks by collin - souffrin & dumont ( 1989 ) and dumont & collin - souffrin ( 1990a , b ) is that , because of their large densities ( of order cm3 ) and column densities ( of order cm2 ) , the disks would emit predominantly low - ionization lines ( e.g. , balmer lines , , and ) rather than high - ionization lines .
thus we have tried to verify that the profiles and strengths of the and h lines are consistent with an accretion disk origin using blue spectra of four objects whose lines are redshifted to wavelengths longer than the atmospheric cutoff ( 3c 17 , pks 034037 , pks 115134 , and cso 643 ; @xmath15 ) . to carry out a more complete and detailed test ,
we have obtained uv spectra of several objects with relatively low redshifts ( @xmath16 ) with the _ hubble space telescope ( hst)_. the results for the first of these objects , arp 102b , have been published ( see 5.2 ) , while a paper presenting our study of several more objects is in preparation . according to the photoionization models we have adopted , the hemissivity is proportional to @xmath12 at moderate radii but it rises steeply towards the center at small radii ( approximately as @xmath17 ) .
the line emissivity is also proportional to @xmath12 at moderate radii and becomes `` shallower '' at small radii .
the emissivity of _ all _ lines saturates at very high incident fluxes ( i.e. , at very small disk radii ) ; the incident flux required to saturate h is about an order of magnitude higher than that for h or .
the emissivity laws described above are summarized in figure 7 of collin - souffrin & dumont ( 1989 ) .
thus , we parameterized the the disk emissivity as a broken power law of the form @xmath18 where @xmath19 is a constant of proportionality , @xmath20 is the break radius at which the power - law index changes , and the factor @xmath21 ensures that the emissivity is continuous at @xmath20 .
the power - law indices were held fixed to @xmath22 and @xmath23 for the h fit .
for the fit we set @xmath23 and allowed @xmath24 to vary between 2 and 3 to simulate the flattening of the emissivity profile at small radii .
the broadening parameter , @xmath10 , and of course the inclination angle of the disk , @xmath5 , were held fixed to the values determined by fitting the h profile in paper i. we also tried to keep the inner and outer radii of the line - emitting disk close to the values determined from fitting h , although this was not always possible ( the parameters determined from fits to the h profiles are more robust because the h profile is not seriously contaminated by strong , narrow lines , and has the simplest emissivity law ) . in figure [ fprof ]
we show a montage of the the h , h , and profiles for each of the three objects , plotted on a common velocity scale , with the best - fitting models superposed .
the model parameters are summarized in table [ tfit ] . in the case of 3c 17
we used an elliptical disk model to describe the line profile ( see [ s_fit_ha ] , above ) ; the model for the h line also describes the low-@xmath25 profile of the weak h line . in the case of cso 643
we applied a blueshift of 1370 to the model profiles ( see [ s_fit_ha ] and [ s_indiv ] ) .
we find that the model parameters expected on theoretical grounds produce acceptable fits to the h profiles .
however , the goodness of a fit can only be judged for the blue half of the line profile because the red side is overwhelmed by the very strong [ ] 4959 , 5007 doublet from the narrow - line region . as a result
, we can not easily distinguished between a simple and a broken power - law emissivity prescription for the h line .
the observed profiles are also consistent with a disk origin , although they appear to come from a larger region of the disk than what we expected on theoretical grounds .
in fact , in some cases we can only obtain a lower limit to the outer radius of the -emitting region of the disk .
the relative strengths of the h , h , and lines of the four objects were measured by integrating the flux in the best - fitting models and they are included in table [ tlrat ] .
this table also includes the h / h ratios of all other double - peaked emitters for which the h lines have been observed and the /h ratio for arp 102b from halpern et al .
( 1996 ) ; these ratios were measured by superposing the two profiles and scaling them so that they matched .
all line ratios have been corrected for reddening in the interstellar medium of the galaxy ( see 5.4 ) .
we find that the measured h / h ratios are generally consistent with the predictions of dumont & collin - souffrin ( 1990a ) , but the measured /h ratios are higher than the model predictions by a factor of about 3 .
the discrepancy can be the result of one or a combination of the following effects : 1 .
the disk is not the only source of emission ; a wind overlaying the geometrically thin disk may contribute significantly to the observed flux but not to the observed h flux .
the profile of the line from the wind is likely to be narrower than the that from the disk and single - peaked , and it could fill in the trough between the two peaks of the disk profile .
thus the situation would be analogous to the lines of arp 102b , where there is strong evidence that the line profiles comprise contributions from two different sources with different dynamics and ionization structure ( halpern et al .
the ionizing flux intercepted by the disk is so low that the metal lines are formed primarily by collisional excitation rather than recombination , while the balmer lines are formed almost exclusively by recombination .
this effect is discussed by collin - souffrin & dumont ( 1989 ) and reiterated by dumont & collin - souffrin ( 1990a ) .
these authors note that such an effect is an important characteristic of weakly - ionized media .
thus , the /h ratio rises above unity in the outer disk , with the consequence that the outer disk makes a significant contribution to the line but not to the h line . this behavior is not captured by the simple power - law parameterization that we have adopted .
our overall conclusion is that the photoionization models of collin - souffrin & dumont ( 1989 ) and dumont & collin - souffrin ( 1990a , b ) fare moderately well against the observations .
they provide emissivity prescriptions that can be used to fit the observed line profiles well , and they can predict the h / h ratios fairly well . we note in closing that as more observations of lines become available , tests of these models can be refined and extended .
following paper i , we repeat and refine the statistical comparisons between the starlight fraction and narrow - line equivalent widths of double - peaked emitters using the combined data from both papers .
we also compare the distribution of the ratios of the narrow oxygen lines , [ ] / [ ] and [ ] / [ ] , of double - peaked emitters and other radio - loud agns .
these line ratios serve as a diagnostic of the ionization state of the narrow - line region . unlike paper
i , we put all double - peaked emitters in the same group , without distinguishing objects whose h profiles are well - described by a disk model ( see our discussion in 3 ) . in table
[ tstar ] we list the _ observed _ equivalent widths of the [ ] and [ ] lines and the fractional contribution of starlight to the optical continuum around h in the newly observed objects . in table [ tlrat ] we list the oxygen line ratios of double - peaked emitters and comparison objects , which were either measured from our own spectra or drawn from a number of sources in the literature .
the effects of galactic reddening were corrected using the reddening measurements of schlegel , finkbeiner , & davis ( 1998 ) and assuming the seaton ( 1979 ) reddening law . to guard against uncertainties in the absolute flux calibration of any individual spectrum ,
line ratios were computed only if both lines could be measured from the same spectrum .
we were able to match the redshift range of the [ ] / [ ] comparison sample to that if the double - peaked emitters ( i.e. , @xmath2 ) . in the case of the [ ] / [ ] comparison sample , however , we had to extend the redshift range to @xmath26 so as to build up a sizable sample from published data .
the results of these comparisons are shown graphically , in the form of histograms in figures [ fstar ] and [ flrat ] .
the average values of the relevant quantities are summarized in table [ tcomp ] along with the probability that their distribution in double - peaked emitters and other radio - loud agns were drawn from the same parent population .
the results of this comparison reaffirm the conclusions of paper i : the mean starlight fraction is in double - peaked emitters is three times higher than in other radio - loud agns , while the mean rest - frame equivalent widths of the [ ] and [ ] lines are 36% and 60% higher .
the oxygen line ratios suggest that the narrow - line regions of double - peaked emitters are at a lower ionization state than than those of typical radio - loud agns .
the oxygen line ratios double - peaked emitters , are , in fact , reminiscent of liners ( low - ionization nuclear emission regions ; after heckman 1980 ) , although only arp 102b fulfills the formal definition of that class .
we discuss this connection further in our interpretation of the properties of double - peaked emitters in 6.1 .
in addition to the properties of the narrow lines we have also studied the properties of the broad lines of double - peaked emitters using a similar approach . in table
[ tlrat ] we list the h / h and /h ratios of double - peaked emitters and comparison objects . measurements , reddening corrections , and construction of comparison samples was carried out as for the narrow - line ratios ( including extension of the redshift range of the /hcomparison sample to @xmath26 ) .
the /h ratios for many of the comparison objects were kindly provided by m. s. brotherton and b. j. wills . in figure [ flrat ]
we compare the distributions of broad - line ratios in double - peaked emitters and in other radio - loud agns , while the mean values and k - s probabilities are included in table [ tcomp ] .
we find that the distributions of the h / h ratio are considerably different , while those of the /h ratio are quite similar , although this is based on a small number of double - peaked emitters . at first glance
, one can interpret the larger h / h ratio of double - peaked emitters as the result of reddening , which would imply a difference in the mean extinction in the two groups of about @xmath27 .
however , this interpretation would be in conflict with the
/h ratios of double - peaked emitters which are not accordingly smaller than those of the comparison objects . since this statement is based only a small number of measurements of the /h ratio in double - peaked emitters , it should be checked again as more data become available . if we take the /h
distributions at face value , we conclude that the broad - line regions of disk - like emitters are not unusually reddened .
rather , it is more plausible that the h / h ratios of double - peaked emitters are enhanced by collisional excitation , which could be a consequence of an origin of the broad balmer and lines in the dense material of an accretion disk .
all of the double - peaked emitters considered here are radio - loud .
the radio properties of the entire set ( 5 ghz monochromatic luminosities , spectral indices , and morphologies ) are listed in table [ tradio ] .
most double - peaked emitters are powerful fr ii radio galaxies ( fanaroff & riley 1974 ) with @xmath28 w hz1 and double - lobed radio morphologies ; three objects are associated with powerful but compact steep - spectrum radio sources .
four objects ( iras 0236.63101 , 1e 0450.31817 , pks 0921213 , and arp 102b ) have @xmath29 w hz1 , which is intermediate between what are commonly considered high and low radio luminosities .
the infrared properties of double - peaked emitters are poorly known .
only two of the nearest and best - studied objects have been detected by _ iras _ : arp 102b and 3c 390.3 . in the broad - band spectral energy distributions ( seds ) of these two objects ( chen et al .
1989 ; golombek , miley , & neugebauer 1988 ) the usual `` uv bump '' is absent .
instead these seds peak in the infrared band around 25 @xmath30 m .
the weakness of the non - stellar continuum in the optical spectra of all double - peaked emitters is probably a manifestation of the absence of a `` uv bump '' .
the uv spectra of double - peaked emitters can afford important tests of models for the origin of the double - peaked lines via the strong , high - ionization , resonance lines that they cover .
the only double - peaked emitter with a published uv spectrum from the @xmath31 ( i.e. , with a moderate - resolution and a high signal - to - noise ratio ) is arp 102b ( halpern et al 1996 ) .
this spectrum shows that is the only uv line which is as broad as the balmer lines and that the twin peaks of the low - ionization lines do not have counterparts in the high - ionization lines . the upper limit to
the ly / h ratio in the twin peaks is 0.15 .
the absence of a double - peaked ly line can not be the result of reddening because this would be in conflict with the observed /h and h / h ratios which are typical of radio - loud agns ( see 5.4 ) .
we discuss the implications of this spectrum further in [ s_disc_tests ] .
we are currently in the process of analyzing @xmath31 spectra of more double - peaked emitters .
a fair fraction of double - peaked emitters are bright enough in the x - ray band to allow their x - ray properties to be studied .
soft x - ray fluxes of most double - peaked emitters are listed in table [ tdisk ] , which is an expanded and updated version of a similar table from paper i. the hard x - ray spectra of many of the nearby double - peaked emitters have been studied with _ asca _
( 0.510 kev ; see , for example , woniak et al . 1998 and sambruna , eracleous , & mushotzky 1999 ) and _ rxte _ ( 4100 kev ;
see eracleous , sambruna , & mushotzky 2000 , and references therein ) .
the general conclusion is that the x - ray spectra of most double - peaked emitters do not differ from those of typical radio - loud agns .
the accretion disk model of chen et al . ( 1989 ) and chen & halpern ( 1989 ) can provide an explanation for the unusual properties of double - peaked emitters .
the predicted line profiles can fit the data and the model also provides a self - consistent framework within which the other of the observational properties of this group of objects can be understood .
motivated by the need to balance the energy budget of the outer , line - emitting part of the disk , chen & halpern ( 1989 ) proposed that the inner accretion disk is a vertically extended structure that can illuminate the outer disk effectively and power the line emission .
the energy budget problem comes about because the hluminosity is uncomfortably close to , or in some cases in excess of , the power dissipated by viscous stresses in the line - emitting portion of the disk .
this power deficit is demonstrated in table [ tdisk ] ( an updated version of table 9 of paper i ) and in figure [ fdisk ] , where we compare their h luminosities with the power output of their line - emitting disks .
the power output of the disk was computed using equation ( 4 ) of paper i and the x - ray luminosity , under the assumption that the x - ray luminosity represents 1% of the available accretion power . in 30% of the cases
the h luminosity is higher than the local power output of the line - emitting portion of the disk and in 90% of the cases this fraction is more than 0.1 ( figure [ fdisk]a ) . on the other hand
, the soft x - ray luminosity always exceeds the h luminosity ( their ratio is more than 10 in 70% of the cases ) .
this comparison suggests very strongly that the balmer lines are powered by photoionization of the outer disk by radiation produced in the inner disk .
the inner accretion disk was hypothesized by chen & halpern ( 1989 ) to be a hot , ion supported torus ( rees et al .
1982 ) which radiates the available accretion power inefficiently , and its existence is favored at low accretion rates ( one to two orders of magnitude below the eddington rate ) .
the preferential association of double - peaked emitters with radio - loud agns was also explained automatically , since the original ion torus model was intended to explain the formation of radio jets .
the inner radius of the line - emitting part of the disk was , therefore , identified with the radius at which the disk puffs up to form the torus , while the outer radius was identified with the radius at which the disk fragments into discrete , self - gravitating clouds .
the structure and properties of the ion torus are very similar to those of advection - dominated or convection - dominated accretion flows ( adafs and cdafs , possibly accompanied by an outflow ) discussed more recently by many authors , including narayan & yi ( 1994 , 1995 ) , blandford & begelman ( 1999 ) , igumenchev & abramowicz ( 1999 , 2000 ) , and narayan , igumenchev , & abramowicz ( 2000 ) .
the sed of an ion torus is relatively hard , lacking a `` uv bump '' ( the hallmark of a geometrically thin accretion disk ) , having instead a shape that resembles a power law in the uv and soft x - ray bands , and a peak in the far - ir , between 30 and 300 @xmath30 m ( see , for example , di matteo et al . 1999 , 2000 , and ball , narayan , & quataert 2001 ) .
such an sed explains the weakness of the non - stellar continuum in the optical spectra of double - peaked emitters in general .
moreover , the ir peak in the seds of arp 102b and 3c 390.3 can be identified with the peak predicted by emission models for an ion torus .
the unusually low ionization state of the narrow - line regions of double - peaked emitters ( see [ s_optical ] ) can also be understood in this context and with the help of photoionization models for liners ( halpern & steiner 1983 ; ferland & netzer 1983 ) .
these models show that an otherwise `` normal '' narrow - line region , illuminated by an ionizing continuum with a power - law spectrum and a low ionization parameter produces emission lines with relative strengths very similar to what is observed in double - peaked emitters .
more recent photoionization calculations , carried out by nagao et al .
( 2002 ) and intended specifically for double - peaked emitters , reach the same conclusion . in particular , nagao et al .
( 2002 ) have compared the narrow - line ratios of double - peaked emitters and of a large collection of other radio - loud agns with photoionization models employing two different seds : one consisting of a simple power law and one including a `` uv bump . ''
they find that the narrow - line ratios of double - peaked emitters are best explained by models with a power - law sed .
the origin of the broad , double - peaked balmer lines in an accretion disk is supported further by tests on two of the best studied objects , arp 102b and 3c 390.3 , which we describe below . in the case of arp 102b ,
the profile of the ly@xmath32 line is single - peaked , bell - shaped , and considerably narrower than those of the double - peaked balmer lines ( see figure 3 of halpern et al .
the best interpretation of this difference is that the region producing the double - peaked balmer lines has such a high density and column density ( @xmath33 and @xmath34 ) that the ly@xmath32 photons it produces are trapped by resonance scattering and destroyed by collisional de - excitation .
in fact , this dramatic difference in line profiles was _ predicted _ by rokaki , boisson , & collin - souffrin ( 1992 ) based on the accretion disk photoionization models of collin - souffrin & dumont ( 1989 ) . a reverberation mapping campaign of 3c 390.3 by the _ international agn watch _ ( dietrich et al .
1998 ; obrien et al .
1998 ) has shown that the balmer lines respond to variations of the continuum with a lag of 20 days and that the blue and red sides of the line respond together with an upper limit on the delay between them of 4 days .
similar results were obtained by sergeev et al .
( 2002 ) , who also estimated the transfer function of the h line of 3c 390.3 and found it to peak sharply at a lag of 60 days .
this is exactly the behavior that one would expect from an accretion disk , as noted by gaskell ( 1999 ) .
moreover , sergeev , pronik , & sergeeva ( 2000 ) find the same behavior in arp 102b , with the broad h line lagging the continuum variations by less than 66 days .
estimates of the mass of the central black hole in 3c 390.3 from reverberation mapping range from @xmath35 ( peterson & wandel 2000 ) to @xmath36 ( sergeev et al . 2002 ) .
finally , we note a reassuring consistency check based on the superluminal motion observed in the radio core of 3c 390.3 by alef et al .
( 1994 , 1996 ) . the combination of a number of constraints on the orientation of the radio jet yields an inclination angle of the jet of 19@xmath3742@xmath37 ( eracleous , halpern , & livio 1996 km s1 mpc1 .
if we take @xmath38 km s1 mpc1 , we obtain an upper limit of 42 on the jet inclination angle . ] ) .
this is in good agreement with the inclination angle of the axis of the accretion disk of @xmath39 degrees obtained in paper i by fitting the hprofile .
the observations described in this section also serve as tests of alternative scenarios for the origin of double - peaked lines , which we discuss in later sections . in [ s_fit_ha ] we noted that the simplest , axisymmetric disk model does not describe the h profiles of pictor a and cso 643 extremely well , which led us to use more sophisticated models . here
we describe the physical motivation for the models that we used .
_ pictor a_. the double - peaked lines appeared some time between 1983 and 1989 .
a spectrum taken in 1983 by filippenko ( 1985 ) shows a single - peaked h line with very broad wings , while spectra taken between 1989 and 1994 ( halpern & eracleous 1994 ; sulentic et al .
1995a ) show the h line to be double - peaked and quite variable .
halpern & eracleous ( 1994 ) speculated that the appearance of the double - peaked line was related to a major structural change in the accretion disk , while sulentic et al .
( 1995a ) attributed the double - peaked lines to a bipolar outflow
. an important feature of the h profile of pictor a is that the red peak is stronger than the blue ( at least at the time that our spectrum was taken ) , contrary to the prediction of the simplest disk models .
this feature , however , does not preclude the origin of the line in an accretion disk .
any departure of the disk from axisymmetry will lead to line profiles that differ from those predicted by the simplest models .
specific examples of non - axisymmetric disks include ( but are not limited to ) disks with spiral waves ( chakrabarti & wiita 1994 ) , eccentric disks ( eracleous et al .
1995 ) , and disks with warps induced by intense irradiation ( pringle 1996 ; maloney , begelman & pringle 1996 ) .
in fact , the dynamical signatures of the first two of these examples of non - axisymmetric disks have been detected in cataclysmic variables ( steeghs , harlaftis , & horne 1997 ; baptista & cataln 2000 ; patterson , halpern , & shambrook 1993 ) .
refining the simplest disk model by introducing an eccentricity allows us to fit the h profile of pictor a , as illustrated in figure [ ffit]c .
such departures form the simplest model are not uncommon , as shown by eracleous et al .
we must also note that the h profile of pictor a has been varying dramatically over the past decade ( see the sequence of spectra presented by eracleous & halpern 1998 ) , which means that the particular models that we discuss here are at best applicable to a single epoch .
_ cso 643_. the double - peaked h profile of cso 643 resembles those of 3c 59 , pks 034037 , and cbs 74 : they all have a pronounced red shoulder , which is much more extended than the blue one , and require small inclination angles ( @xmath40 ) for a good fit .
however , in the cso 643 the net redshift of the entire profile is smaller than in the other objects preventing us from obtaining a good fit with a simple circular - disk model .
there are two ways to improve the fit : ( a ) we can apply an _ ad hoc _ blueshift to the model of 1370 . the best - fitting circular disk model found under this assumption
is shown in figure [ ffit]e overlayed on the observed profile .
the model parameters are summarized in table [ tfit ] .
( b ) we can use a more sophisticated model with more free parameters , such as an elliptical disk model ( eracleous et al .
1995 ) . the best - fitting model of this type is shown in figure [ ffit]g and the model parameters are summarized in table [ tfit ] . either of the above modifications to the simple disk model can be physically justified by appealing to the effects of an unseen supermassive companion to the accreting black hole ( see , for example , rees et al .
1982 ) . in case
( a ) the blueshift is a result of the orbital motion of the accreting black hole around the center of mass of the supermassive binary . in case ( b ) the eccentricity of the hypothesized elliptical disk is excited by the tidal effect of the unseen companion ( see the discussion in eracleous et al .
1995 and references therein ) .
in addition to accretion - disk emission three other scenarios have been suggested for the origin of double - peaked emission lines , namely ( a ) emission from a binary broad - line region associated with a supermassive binary black hole , ( b ) emission from the oppositely directed sides of a bipolar outflow , and ( c ) emission from a spherically symmetric broad - line region illuminated by an anisotropic source of ionizing radiation . in this section we describe these scenarios and evaluate them in the light of the currently available data .
we consider all of the observational consequences of each scenario and we assess it based on how well it can explain the observational properties of double - peaked emitters .
the idea that supermassive binary black holes ( resulting from the merger of their parent galaxies ) reside in the nuclei of radio - loud agns was suggested by begelman , blandford , & rees ( 1980 ) , who also noted that double - peaked emission lines could be an observational consequence of such a scenario .
gaskell ( 1983 , 1988 , 1996a , b ) identified a number of radio - loud agns with displaced broad emission lines as candidate supermassive binaries and proposed that supermassive binaries are very common in both radio - loud and radio - quiet agns with double - peaked emitters being the most obvious cases . in this scenario
the spectroscopic properties of the double - peaked emitters should not differ from the those of the average ( radio - loud ) agn , which is in contradiction with the observational properties described in 5 and [ s_disc_tests ] ( most notably with the dramatic difference between the profiles of the h and ly lines of arp 102b and the reverberation of the hline of 3c 390.3 ) .
another serious disagreement between this scenario and the observations results from its prediction that the two peaks of a double - peaked line should drift in opposite directions as a result of the orbital motion of the binary .
although the expected orbital periods can range from decades to centuries , the signature of orbital motion can be detected , in spectra spanning one or two decades .
thus halpern & filippenko ( 1988 , 1992 ) searched for radial velocity variations in the displaced peaks of arp 102b and 3c 332 , using spectra spanning a decade , but did not find any . a more extensive study of arp 102b , 3c 390.3 , and 3c 332 by eracleous et al .
( 1997 ) did not find any evidence for orbital motion ; a suggestive variability trend found by gaskell ( 1996a ) in 3c 390.3 lasted only until 1988 ( eracleous et al . 1997 ; shapovalova et al . 2001 ; sergeev et al .
2002 ) the lack of evidence for orbital motion yielded lower limits on the binary masses of , which are in conflict with a number of other observations and with theory , leading to the rejection of this scenario as a general explanation for double - peaked emission lines ( see eracleous et al .
1997 for a more detailed discussion ) .
since double - peaked emitters are radio - loud agn it is conceivable that the double - peaked lines arise in a bipolar outflow resulting from the interaction of the powerful radio jets with gas immediately around the central engine .
norman & miley ( 1984 ) discussed the interaction of the jets with the emission - line regions in qualitative terms , while zheng , binette , & sulentic ( 1990 ) constructed quantitative models for the profiles of emission lines produced in the outflow .
such models were applied to the double - peaked h lines of 3c 390.3 and pictor a by zheng , veilleux & grandi ( 1991 ) and by sulentic et al .
( 1995a ) , respectively .
bipolar outflow models are not immediately appealing , however , because statistical studies of the h profiles of radio - loud agns suggest that the line - emitting gas is arranged in a flat disk perpendicular to the radio axis ( see , for example , wills & browne 1986 ; jackson & browne 1991b ; brotherton 1996 ; marziani et al . 1996 ) . the conclusions of these studies and the fact that most double - peaked emitters are associated with double - lobed radio sources argue against bipolar outflows and contradict the suggestion by sulentic et al .
( 1995a ) that double - peaked emitters represent an extreme segment of the radio - loud agn population in which the axis of the outflow is oriented very close to the line of sight .
the following additional observational constraints make bipolar outflow models for double - peaked emission lines unlikely . 1 .
the dramatic difference between the observed ly and h profiles of arp 102b ( halpern et al 1996 ) can not be easily explained .
the radial velocity gradient in the outflow should reduce the optical depth of ly photons and allow them to escape easily with the consequence that ly and the balmer lines should have similar profiles .
reverberation mapping of 3c 390.3 ( see [ s_disc_tests ] ) has shown that both sides of the h line respond together to changes in the ionizing continuum , contrary to the expectation that the blue - shifted side of the line should respond first , with hardly any lag from the continuum variations .
if the double - peaked lines of 3c 390.3 originate in a bipolar outflow , the reverberation results reviewed in [ s_disc_tests ] imply that the outflow should be viewed nearly at right angles to its axis ( @xmath41 , following livio & pringle 1996 and livio & xu 1997 ) . however , this orientation is in contradiction with the observed radio properties of 3c 390.3 ( see discussion in [ s_disc_tests ] ) .
this model was discussed in its general form by goad & wanders ( 1996 ) although it was originally considered by wanders et al .
( 1995 ) as an interpretation of the reverberation mapping results for the seyfert galaxy ngc 5548 .
the broad - line region is assumed to consist of a large number of clouds in randomly inclined keplerian orbits occupying a thick spherical shell .
the clouds are photoionized by a central source ( presumably an accretion disk ) which emits anisotropically . in the specific picture discussed by goad & wanders ( 1996 )
the anisotropic emission was described by two conical beams which could possibly be superposed on an isotropic `` background '' illumination . for specific combinations of the beam opening angle and orientation of the observer relative to the axis of the beam , the resulting emission - line profiles appear double - peaked and the expected frequency of double - peaked lines is consistent with what is observed .
it was also suggested that the uv spectrum of arp 102b could be explained in this context if the ionizing continuum consists of a hard x - ray power - law emitted isotropically and a softer , `` uv bump '' spectrum making up the two conical beams ( cf , netzer 1987 ) .
moreover , high- and low - ionization lines would , in general , have different profiles .
thus , an observer whose line of sight is within the cone of the uv beam would find the the ly and lines to be relatively narrow and single peaked , but the balmer and lines would appear very broad and double peaked , just as observed in arp 102b .
the weakness of the non - stellar continuum in the optical spectra of double - peaked emitters could also be explained by placing the observer s line of sight outside of the cones of the uv beams .
although this scenario has its attractions , its applicability to double - peaked emitters does not withstand close scrutiny .
it is at odds with the observations in the following ways : 1 . it does not provide a self - consistent explanation of all of the properties of arp 102b
: in order to explain the difference between the profiles of the ly and balmer , the observer s viewing direction must lie within the conical uv beams , but to explain the weakness of the non - stellar continuum relative to the starlight , the observer s viewing direction must be outside the conical uv beams .
2 . it does not explain the unusually strong low - ionization narrow lines of double - peaked emitters . from a theoretical perspective
the physical basis for the assumed cloud distribution and kinematics is questionable .
even if such a system of clouds on randomly inclined , keplerian orbits could be produced at all , it would be destroyed by collisions in 100 dynamical times , or in 200 years in the case of 3c 390.3 ( see the general estimate and discussion in matthews & capriotti 1985 ) . an additional , and arguably more catastrophic , physical effect is the decay of the orbits of discrete clouds due to a poynting - robertson drag , which could occur on a time scale of 660 years ( see mathews & capriotti 1985 ) .
we therefore doubt that this picture of the broad - line region is physically realizable . if the h emission line profiles of approximately 20% of radio - loud agns are to be attributed to accretion disks , what of the remaining objects ?
our understanding of the dynamics of the line - emitting gas would improve tremendously if a single model could be found that would describe the majority of emission - line profiles , especially if the same model can explain double - peaked profiles .
as such , double - peaked emission lines place important constraints on any model that seeks to explain agn broad - line regions in a universal way .
a possible new clue is the discovery of double - peaked balmer emission lines in many liners , which underscores the connection between the two types of hosts .
examples of liners with double - peaked emission lines include ngc 1097 ( storchi - bergmann et al .
1993 ) , m81 ( bower et al .
1997 ) , ngc 4203 ( shields et al .
2000 ) , ngc 4450 ( ho et al . 2000 ) , and ngc 4579 ( barth et al .
the prototypical double - peaked emitter , arp 102b is also a liner ( stauffer , schild , & keel 1983 ) .
since liners are found in about 30% of all nearby galaxies and a significant fraction of them could be agns , the incidence of double - peaked emission lines among the entire agn population could be quite high .
the exact fraction of agns among liners and the frequency of double - peaked emission lines in liners remain to be quantified , however . motivated by these considerations
, we explore here whether accretion disk models , which offer the best explanation for double - peaked lines , can also explain the single - peaked profiles observed in the majority of agns . in support of this approach , we note that our comparison of the widths and shifts of double - peaked and single - peaked emission lines suggests a similarity between the corresponding broad - line regions . furthermore
, if we attribute the mean redshift measured over all radio - loud agns to the combined effects of gravity and transverse motions of the line - emitting gas in a keplerian disk , it allows us to infer a characteristic distance of the line production site from the central object of @xmath42 , comparable to the outer radii of the line - emitting disks found in 4 and in paper i. this estimate supports the above suggestion that all radio - loud agns may harbor line - emitting accretion disks and/or additional line - emitting gas in related structures ( e.g. , an accretion - disk wind ) .
the following possibilities for the origin of single - peaked lines in accretion disks have been discussed in the literature : 1 . the line - emitting disk in most agns could be quite large with a ratio of outer to inner radius of order 10 or more .
this would bring the two peaks of a double - peaked line close together and make the profiles appear single - peaked ( dumont & collin - souffrin 1990a ; jackson , penston , & prez 1991 ) . to illustrate this point we have simulated h spectra by assuming that the broad h line is produced in a large accretion disk , namely one with an inner radius of order a few hundred @xmath43 and an outer radius
greater than @xmath44 i.e. , @xmath45 . to complete the simulation we added the usual narrow lines in the vicinity of h and poisson noise .
the narrow lines were assumed to have gaussian profiles with a fwhm of a few hundred , while the noise was generated assuming a @xmath25 of 30200 in the continuum .
we explored a range of inclination angles between 10 and 40 and a range of emissivity power - law indices between 1.5 and 3 . in figure [ fsimul ]
we show a montage of simulated spectra spanning a range of parameter values .
these examples were chosen because they resemble very closely some of the observed h spectra presented in paper i. thus , large accretion disk models can reproduce , at least qualitatively , a wide variety of profiles observed in radio - loud agns . moreover , red asymmetries are quite common , particularly below half maximum of the profile , which is a desirable feature since this is commonly observed . the large outer radius of the line emitting disk and the h+ [ ] narrow - line complex can conspire to hide the characteristic twin shoulders that are associated with lines from a disk .
we find that `` shallow '' emissivity profiles ( described by power - law indices @xmath46 ) produce line profiles that bear a closer resemblance to the observed ones than those produced by `` steep '' ( @xmath47 ) emissivity laws . in the context of the photoionization calculations of collin - souffrin & dumont ( 1989 )
this can be understood as the result of a very luminous central source .
2 . the disks in most agns could be oriented close to face on .
this possibility was explored by corbin ( 1997a ) who computed the profiles resulting from nearly face - on , flattened broad - line regions resembling face - on disks .
some of the simulated h profiles that we present in figure [ fsimul ] also correspond to face - on disks .
the main conclusion from these calculations is that face - on disk models reproduce many of the desired properties of observed line profiles , including a single peak , extended red wings ( i.e. , redward asymmetries ) , and a net redshift of the entire line .
double - peaked line profiles from a disk can easily be turned into single - peaked profiles by the presence of a disk wind .
murray & chiang ( 1997 ) have shown that although the outflow velocity of the wind is much smaller than the rotational velocity , its velocity gradient is as large as the rotational velocity gradient .
the main physical consequence is that photons can escape much more easily along lines of sight with a small projected velocity .
the resulting line profiles are single peaked with broad wings even though the emission comes from gas that is essentially on circular orbits .
this was illustrated clearly by murray & chiang ( 1997 ) who computed profiles of uv resonance lines arising in a wind and compared them to observed line profiles of quasars .
an additional illustration was provided by chiang & murray ( 1996 ) who showed that the observations of reverberation in the line of ngc 5548 could also be explained , at least qualitatively , in the context of an accretion - disk wind model . of the possibilities discussed above
, we favor the accretion - disk wind scenario because of a number of additional appealing features that it possesses , as we discuss further below .
in particular , it offers a way of connecting double - peaked emitters to the greater agn population .
it also provides an explanation for the h blueshifts observed in some double - peaked emitters ( see figure [ fshifts ] and 3 ) , since a low - optical depth wind can impart a blueshift on lines from the disk without altering their double - peaked profiles .
we therefore propose that the wind is the broad - line region in most agns , which accrete at rates that are a sizeable fraction of the eddington rate .
this idea is by no means new ; it was suggested , for example , by murray et al .
( 1995 ) to explain broad - absorption line quasars and by elvis ( 2000 ) to explain the absorption features observed in the x - ray spectra of quasars .
the dynamics of the wind were worked out analytically by murray et al .
( 1995 ) and through detailed numerical simulations by proga , stone , & drew ( 1999 ) and proga , stone , & kallman ( 2000 ) , which confirm the general analytic results . to show how double - peaked emitters and low - luminosity agns fit into this scheme
, we note that the structure of the accretion flow is controlled largely by the eddington ratio , i.e. , the ratio of the accretion rate to the eddington rate ( @xmath48 ) . at large eddington ratios ( @xmath49 )
the inner accretion disk is geometrically thin and optically thick , as described by the model of shakura & sunyaev ( 1973 ) .
its emitted sed has a prominent `` uv bump , '' which can exert a substantial radiation pressure on the atmosphere of the outer disk via resonance line absorption , accelerating a wind .
we posit that such a structure applies to seyfert galaxies and quasars . at the opposite extreme of a low eddington ratio ( @xmath50 ) the inner disk changes to an ion torus whose emitted
sed is a hard x - ray power - law without a `` uv bump '' .
when such a hard spectrum impinges on the atmosphere of the outer disk it ionizes it and reduces the potential of exerting radiation pressure via resonance line absorption ( see , for example , the discussion by murray et al .
the potential for a significant radiation pressure is reduced further by the lack of substantial numbers of uv photons in the incident sed .
thus the wind becomes feeble and the accretion disk is unveiled , allowing the observer to see double - peaked emission lines from the disk proper . in such objects
the feeble wind would be the primary source of uv resonance emission lines , which would be single peaked as a result of their origin ( see the discussion by collin - souffrin & dumont 1989 ) .
moreover , the wind may also be observable through absorption lines that it produces .
such wind - like absorption lines have actually been detected in arp 102b ( halpern et al .
1996 ; eracleous , halpern , & charlton 2003 ) and other double - peaked emitters ( eracleous 2002 ) .
we have presented the completion of our survey of moderate redshift , radio - loud agns whose primary motivation was to search for more examples of double - peaked emission lines .
we find that 20% of the objects surveyed have h lines with double peaks or twin shoulders . in 17 of the 24 cases the blue peak / shoulder is stronger than the red .
the h profiles of 13 of these 17 objects can be fitted quite well by the simplest possible disk model ( homogeneous and axisymmetric ) .
the h profiles of many of the remaining objects can be fitted by more sophisticated disk models ( non - axisymmetric and/or inhomogeneous ) .
double - peaked emitters possess a number of additional properties which distinguish them from the average radio - loud agn and make them similar to accretion - powered liners .
a consideration of all of the available data shows an origin of double - peaked emission lines in an accretion disk to be the preferred interpretation .
alternative scenarios are rather unsatisfying : although they can produce double - peaked emission lines , at least in principle , they can not explain all of the properties of double - peaked emitters , they fail some direct observational tests , and in some cases their physical foundations do not withstand close scrutiny .
a physical model in which the inner accretion disk is an ion torus that illuminates the outer disk can explain the observed properties of double - peaked emitters .
this model also provides a framework in which the connection between double - peaked emitters and the greater agn population can be sought .
more specifically , we propose that in most agns the broad - line region is an accretion - disk wind , while in double - peaked emitters and other low - luminosity agns ( such as some liners ) the broad - line region is the outer accretion disk .
this transformation is brought about by a decline of the accretion rate from a sizable fraction of the eddington rate to values that are more than two orders of magnitude lower . at these extremely low accretion rates
the structure of the inner disk changes to an ion torus and the wind that cloaks the outer disk diminishes .
this scenario can be tested by comparing the profiles of uv resonance lines with the profiles of the balmer lines of double - peaked emitters .
even though our survey targeted radio - loud agns so as to maximize the efficeincy of finding double - peaked emitters , radio - quiet agns also host double - peaked emission lines . as we noted in [ s_disc_single ]
a significant number of liners have recently been found to possess double - peaked balmer lines .
liners have been traditionally regarded as radio - quiet objects but it was recently argued by ho et al .
( 2000 ) and terashima & wilson ( 2003 ) that the ratio of the radio luminosity of liners to their non - stellar optical luminosity is considerably higher than that of radio - quiet agns . the much larger sample of 116 double - peaked emitters found in the sloan digital sky survey ( sdss
; see strateva et al .
2003 ) comprises 76% radio - quiet agns , indicating that radio - quiet double - peaked emitters are more numerous than radio - loud ones .
however , double - peaked emitters are 1.6 times more likely to be found among radio - loud agns .
finally , although it is beyond the scope of this paper , variability of double - peaked profiles deserves discussion .
we distinguish two types of variability ( a ) short - term variability of the line flux due to reverberation of the ionizing continuum ( time scales of order weeks ) , and ( b ) long term variability of the line profiles ( time scales of order months to years ) . in this paper
we have made extensive use of the results of reverberation mapping of 3c 390.3 ( dietrich et al .
1998 ; obrien et al .
1998 ; sergeev et al .
2002 ) , because their implications are straightforward and important . since
such experiments have been performed only for 3c 390.3 , we have taken these results to be representative of all double - peaked emitters .
however , reverberation mapping of additional objects is sorely needed to check their behavior and to measure the dimensions of the line - emitting region .
studies of long - term profile variations of the line profiles have been carried out for a handful of double - peaked emitters , namely 3c 390.3 , ( e.g. , zheng et al . 1991 ; eracleous et al .
1997 ; gilbert et al .
1999 ; sergeev et al .
2002 ) arp 102b ( e.g. , miller & peterson 1990 ; newman et al .
1997 ) , ngc 1097 ( storchi - bergmann et al .
1995 , 1997 , 2003 ) , 3c 332 ( eracleous et al . 1997 ; gilbert et al .
1999 ) , and profile variability results on pictor a have also been reported ( halpern & eracleous 1994 ; sulentic et al .
1995a ; eracleous & halpern 1998 ) but no comprehensive study exists yet .
the long - term variability patterns show a wide variety , although the most pronounced and obvious pattern is the slow and systematic modulation of the relative strengths of the two peaks . in this paper
we have appealed to profile variations primarily to test the binary black hole hypothesis . however , the interpretation of the long - term profile variations is not yet clear and they do not yet constitute a straightforward test of scenarios for the origin of the lines .
such slow variations are certainly not the result of reverberation of a variable ionizing continuum since the light - crossing time of the line - emitting region is of order a few weeks ( see [ s_disc_tests ] ) .
more generally , reverberation mapping of seyfert galaxies has shown that although the _ integrated fluxes _ of broad emission lines do vary in response to a variable ionizing continuum , the line profiles do not ( e.g. , ulrich et al .
1991 ; wanders & peterson 1996 ; kassebaum et al .
the time scales on which the _ line profiles _ change significantly are longer than the dynamical time of the line - emitting region by a factor of several .
now that accretion disk emission is emerging as the favorite scenario for the origin of double - peaked emission lines , we will be able to investigate the origin of the observed profile variations through detailed case studies and exploit them to learn about dynamical phenomena in agn accretion disks .
we thank the anonymous referee for helpful comments and suggestions .
we are grateful to m. s. brotherton and b. j. wills for allowing us to use their measurements of /h ratios of many radio - loud agns prior to publication .
we also thank a. v. filippenko for obtaining some of the spectra presented in this paper at lick observatory and we acknowledge the help of a. barth in the reduction of these spectra . during the early stages of this work m.e .
was based at the university of california , berkeley , and was supported by hubble fellowship grant hf-01068.01 - 94a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa under contract nas 5 - 26555 .
gaskell , c. m. 1999 in structure and kinematics of quasar broad line regions , asp conf .
, 175 , eds . c. m. gaskell , w. n. brandt , m. dietrich , d. dultzin - hacyan , & m. eracleous ( san fransisco : asp ) , 157 gilbert , a. m. , eracleous , m. , filippenko , a. v. , & halpern , j. p. 1999
, in structure and kinematics of quasar broad line regions , asp conf .
, 175 , eds . c. m. gaskell , w. n. brandt , m. dietrich , d. dultzin - hacyan , & m. eracleous ( san fransisco : asp ) , 189 llllrll 3c 17 & 18.0 & 0.220 & 1994 aug 5 & 1800 & lick 3 m & h , h , + & & & 1996 oct 11 & 3600 & lick 3 m & h , h , + mrc 0041 + 11 & 19.0 & 0.226 & 1994 aug 5 & 600 & lick 3 m & + 4c 31.06 & 18.0 & 0.373 & 1993 dec 13 & 2000 & kpno 4 m & ( b2 0154 + 31a ) + pks 034037 & 18.6 & 0.285 & 1994 feb 15 , 17 & @xmath51 & ctio 4 m & h , h , + 3c 93 & 19.2 & 0.357 & 1993 dec 13 & 10800 & kpno 4 m & + 3c 111 & 18.0 & 0.049 & 1993 dec 13 & 3200 & kpno 4 m & + pks 051148 & 19.0 & 0.306 & 1994 feb 15 & 4800 & ctio
4 m & narrow lines + pictor a & 16.2 & 0.035 & 1994 feb 17 & 3000 & ctio 4 m & + cbs 74 & 16.0 & 0.092 & 1998 jan 30 & 3600 & mdm 2.4 m & + pks 0921213 & 16.5 & 0.053 & 1995 mar 24 & 6600 & kpno 4 m & + 4c 72.16 & 17.9 & 1.462 & 1993 dec 13 & 2700 & ctio 4 m & new redshift + 4c 36.18 & 18.0 & 0.392 & 1993 dec 13 & 7200 & kpno 4 m & + pks 115134 & 17.8 & 0.258 & 1994 feb 15 , 16 & 2700 + 3600 & ctio 4 m & h , h , + pks 133512 & 18.5 & 0.539 & 1994 feb 15 & 2700 & ctio 4 m & new redshift + pks 134611 & 18.0 & 0.341 & 1994 feb 15 & 3000 & ctio 4 m & + cso 643 & 16.7 & 0.276 & 1997 jun 9 & 2700 & kpno 2.1 m & h + & & & 1998 apr 9 & 5400 & mdm 2.4 m & h + & & & 2000 may 31 & 1800 & mdm 2.4 m & h , + pks 145137 & 16.7 & 0.314 & 1994 feb 15 & 1800 & ctio 4 m & + pks 1514 + 00 & 15.6 & 0.053 & 1995 jun 4 & 3600 & kpno 2.1 m & + 3c 351 & 15.3 & 0.372 & 1994 jul 4 & 3600 & kpno 2.1 m & + 3c 381 & 17.2 & 0.161 & 1994 jul 2 , 3 & @xmath52 & kpno 2.1 m & narrow lines + pks 2247 + 14 & 16.9 & 0.235 & 1993 dec 12 & 1407 & kpno 4 m & + pks 2305 + 18 & 17.5 & 0.313 & 1993 dec 12 & 1533 & kpno
4 m & + pks 2312319 & 18.5 & 1.322 & 1994 aug 5 & 1800 & lick 3 m & new redshift + 3c 456 & 18.5 & 0.233 & 1993 dec 13 & 3600 & kpno 4 m & narrow lines + mrc 2328 + 167 & 18.3 & 0.280 & 1993 dec 13 & 2400 & kpno 4 m & + lrrcrrrr mc 0041 + 11 & 11.9 & 4.9 & 0.28 & 8500 & 4200 & 0.0003 & 0.0016 + 4c 31.06 & 5.4 & 23.3 & 0.11 & 18000 & 9000 & 0.0049 & 0.0001 + 3c 111 & 7.7 & 6.3 & 0.04 & 18400 & 4800 & 0.0037 & 0.0010 + pictor a & 55.7 & 30.7 & 0.14 & 29000 & 18400 & 0.0001 & 0.0011 + cbs 74 & 7.7 & 14.4 & 0.17 & 22100 & 9200 & 0.0014 & 0.0035 + pks 092123 & 5.3 & 6.2 & 0.65 & 21900 & 8300 & 0.0015 & 0.0002 + pks 134611 & 4.2 & 5.2 & 0.00 & 11500 & 2300 & 0.0009 & 0.0003 + cso 643 & 10.1 & 26.6 & 0.28 & 21000 & 9000 & 0.0009 & 0.0022 + pks 145137 & 2.6 & 4.9 & 0.00 & 15400 & 3800 & 0.0044 & 0.0003 + pks 151400 & 10.8 & 15.1 & 0.76 & 8300 & 4300 & 0.0022 & 0.0001 + pks 2247 + 14 & 0.0 & 4.4 & 0.00 & 8400 & 3500 & 0.0027 & 0.0013 + pks 2305 + 18 & 14.6 & 15.8 & 0.56 & 8800 & 4400 & 0.0006 & 0.0013 + pks 2328 + 167 & 0.0 & 6.3 & 0.37 & 8800 & 3200 & 0.0017 & 0.0003 + lccc fwhm of h line ( ) & 12,700 & 62,00 & @xmath53 + fwzi of h line ( ) & 21,700 & 17,000 & @xmath54 + starlight fraction & 0.33 & 0.11 & @xmath55 + rest ew of [ ] ( ) & 6.8 & 5.0 & @xmath56 + rest ew of [ ] ( ) & 12.3 & 7.7 & @xmath57 + [ ] / [ ] ratio & 0.19 & 0.07 & @xmath58 + [ ] / [ ] ratio & 0.35 & 0.21 & @xmath56 + broad h@xmath32/h@xmath59 ratio & 5.23 & 4.26 & @xmath60 + broad /h@xmath59 ratio & 2.62 & 2.19 & @xmath61 + llllcllll 3c 17 ( elliptical ) @xmath62 & 360@xmath63 & 800@xmath64200 & 30@xmath644 & 1.7 & & 1400@xmath64200 & 100@xmath65 & [email protected] + 4c 31.06 & @xmath66 & @xmath67 & @xmath68 & 3.0 & & 1200@xmath64200 & & + pks 034037@xmath69 & 210@xmath70 & 1900@xmath64600 & 18@xmath641 & 3.0 & & 1400@xmath71 & & + pictor a ( circular ) @xmath72 & 1000@xmath73 & 1400@xmath71 & 70@xmath74 & 3.0 & & 1000@xmath64200 & & + pictor a ( elliptical ) @xmath75 & 220@xmath76 & 550@xmath77 & 30@xmath78 & 1.5 & & 1000@xmath64200 & 110@xmath79 & 0.29@xmath80 + cbs 74 & 170@xmath81 & 1800@xmath82 & 16@xmath78 & 3.0 & & 1000@xmath83 & & + pks 0921213 & @xmath84 & @xmath85 & @xmath86 & 1.5 & & 600@xmath64100 & & + pks 115134@xmath87 & 300@xmath88 & 2000@xmath89 & 28@xmath90 & 3.0 & & 1600@xmath91 & & + cso 643 ( blueshifted)@xmath92 & 210@xmath93 & 2700@xmath94 & 17@xmath95 & 3.0 & & 1000@xmath64200 & & + cso 643 ( elliptical)@xmath96 & 1600@xmath64800 & 10000@xmath645000 & 55@xmath97 & 3.0 & & 1200@xmath64200 & 90@xmath6440 & 0.5@xmath98 + 3c 17 ( elliptical ) @xmath62 & + pks 034037 & 220@xmath99 & @xmath100 & & 5.1 , 3.0 & 450@xmath64100 & & & + pks 115134 & 400@xmath101 & 2000@xmath64500 & & 5.1 , 3.0 & 1000@xmath102 & & & + cso 643 ( blueshifted)@xmath92 & + 3c 17 ( elliptical ) & + pks 034037 & 200@xmath76 & 2500@xmath103 & & 2.2 , 3.0 & 700@xmath104 & & & + pks 115134 & 170@xmath105 & @xmath106 & & 2.2 , 3.0 & 3000@xmath641000 & & & + cso 643 ( blueshifted)@xmath87 & 200@xmath99 & 3000@xmath641000 & & 2.2 , 3.0 & 500@xmath104 & & &
+ llllllll + 3c 17 & 0.43 & 0.74 & & 7.19 & 1.34 & & 1 + 4c 31.06 & 0.048 & 0.40 & & 4.90 & & & 1 + 3c 59 & 0.066 & 0.15 & & 4.64 & & & 3 + iras 0236.63101 & 0.10 & & & & & & 1,2 + pks 034037 & 0.083 & 0.27 & & 5.28 & 1.91 & & 1 + 3c 93 & 0.055 & 0.26 & & 4.38 & & & 1 + ms 0450.31817 & 0.46 & 0.70 & & & & & 1 + pictor a & 0.58 & 0.50 & & 6.65 & & & 1 + b2 0742 + 31 & 0.028 & & & 5.16 & & & 1 + cbs 74 & 0.087 & 0.18 & & 3.00 & & & 1 + pks 085719 & 0.16 & & & 4.78 & & & 1 + pks 0921213 & 0.22 & 0.11 & & 4.59 & & & 1 + 4c 36.18 & 0.11 & 0.22 & & 5.63 & & & 1 + pks 115134 & 0.10 & 0.33 & & 7.32 & 4.06 & & 1 + cso 643 & & 0.071 & & 3.00 & 3.15 & & 1 + 3c 332 & 0.14 & 0.16 & & 7.38 & & & 3 + arp 102b & 0.78 & 1.2 & & 4.41 & 1.83 & & 4 + pks 1739 + 18c & 0.15 & & & 5.27 & & & 1 + 3c 382 & 0.070 & 0.15 & & 4.82 & & & 5 + 3c 390.3 & 0.064 & 0.070 & & 5.84 & & & 5 + pks 191445 & 0.12 & & & 5.10 & & & 1 + + 3c 18 & & 0.16 & & & & & 10 +
mrc 0041 + 11 & 0.13 & 0.14 & & 5.20 & 1.43 & & 1 + txs 0042 + 101 & 0.42 & & & & & & 8 + 3c 48 & & 0.17 & & 6.76 & 1.10 & & 6,7 + 3c 47 & & 0.13 & & & 2.50 & & 8,9 + pks 020276 & 0.053 & & & 4.21 & & & 1 + pks 0214 + 10 & 0.039 & & & 4.90 & & & 1 + 3c 67 & & 0.31 & & & & & 11 + pks 0403132 & & 0.20 & & & 1.70 & & 8,9 + 3c 111 & 0.076 & & & 2.94 & & & 1 + 3c 120 & 0.059 & 0.047 & & 4.98 & & & 10,12 + 3c 147 & & 0.46 & & & & & 8 + pks 0736 + 01 & & 0.093 & & 3.37 & 1.78 & & 10,6,13 + 3c 206 & & 0.097 & & 4.73 & & & 8,14 + 4c 05.38 & 0.048 & & & & & & 1 + 3c 215 & & 0.17 & & & 1.40 & & 15,9 + pks 0925203 & 0.038 & & & 4.69 & & & 1 + 3c 227 & 0.072 & 0.081 & & 6.37 & & & 5 + 4c 09.35 & 0.035 & & & & & & 1 + 3c 232 & & 0.32 & & & 7.74 & & 13 + 3c 234 & 0.013 & 0.075 & & & & & 3,6 + pks 1004 + 13 & & & & 4.14 & & & 16 + b2 1028 + 31 & 0.031 & & & 3.85 & & & 8,6 + 3c 246 & 0.042 & & & 4.87 & & & 1 + 4c 61.20 & & 0.30 & & & & & 13 + 3c 249.1 & & & & 3.91 & & & 16 + pks 110132 & 0.044 & & & 4.20 & & & 1 + pks 1103006 & & & & & 1.60 & & 9 + pks 1136135 & & 0.083 & & & & & 10 + on 029 & & & & 3.73 & & & 8 + b2 1223 + 25 & 0.020 & & & 3.54 & & & 1 + 3c 273 & & 0.16 & & 4.26 & 0.35 & & 10,13 + pks 123324 & 0.016 & & & 5.68 & & & 1 + 3c 277.1 & 0.35 & 0.081 & & 3.90 & 2.30 & & 8,11,9 + pks 1302102 & 0.080 & & & 3.18 & & & 16
+ 3c 287.1 & 0.24 & 0.74 & & & & & 3 + pks 134611 & 0.054 & & & 3.07 & & & 1 + b2 1351 + 26 & 0.061 & & & 3.56 & & & 16 + pks 135541 & 0.13 & 0.13 & & 4.53 & & & 1,10 + mkn 668 & 0.036 & & & & & & 1 + pks 141719 & 0.033 & 0.046 & & 4.83 & & & 3 + pks 142138 & 0.066 & & & 3.54 & & & 1 + b2 1425 + 26 & 0.074 & & & 6.64 & & & 16 + pks 145137 & 0.040 & & & 4.04 & & & 1 + 4c 69.18 & 0.063 & & & 4.14 & & & 16 + pks 151008 & & 0.17 & & & 2.23 & & 1,10,6 + 4c 37.43 & 0.028 & & & 4.21 & & & 1 + pks 1514 + 00 & 0.50 & 1.2 & & & & & 1 + lb 9743 & 0.007 & & & 4.19 & & & 16 + 3c 323.1 & 0.025 & & & 2.98 & 1.80 & & 16,9 + 3c 334 & & & & & 2.70 & & 9 + 3c 345 & & & & & 3.00 & & 9 + 4c 18.47 & & & & 3.75 & & & 16 + 4c 61.34 & & 0.69 & & & & & 8 + 3c 351 & 0.015 & 0.087 & & & 1.97 & & 6 + b2 1719 + 35 & 0.059 & & & 3.45 & & & 1,16 + 4c 34.47 & 0.070 & 0.058 & & 4.44 & 2.58 & & 6 + pks 1725 + 044 & 0.030 & & & 3.42 & & & 16 + mrc 1745 + 16 & 0.043 & & & 3.93 & & & 1 + 3c 381 & 0.053 & 0.23 & & & & & 3 + 4c 73.18 & 0.030 & & & 3.87 & & & 8,6 + pks 212812 & & 0.088 & & & & & 10 + pks 213514 & 0.092 & 0.090 & & 4.48 & & & 10,16 + pks 2140048 & 0.054 & & & 3.33 & & & 1 + ox 169 & 0.025 & & & 3.23 & & & 1,8 + 4c 31.63 & & & & 3.48 & 2.30 & & 8,9 + pks 220813 & 0.071 & & & 3.90 & & & 1 + 3c 445 & 0.042 & 0.078 & & 6.63 & & & 5 + pks 2227399 & 0.058 & & & & & & 1 + pks 2247 + 14 & 0.018 & & & 5.00 & & & 1 + 4c 11.72 & & 0.13 & & 4.90 & 1.00 & & 8,9 + pks 230271 & 0.046 & & & 3.93 & & & 1 + pks 2305 + 18 & 0.046 & & & 3.98 & & & 1,16 + pks 2345167 & & 0.20 & & & & & 8 + pks 234901 & 0.038 & 0.28 & & 4.07 & & & 3 + lcrll 3c 17 & @xmath107 & 0.62 & fragmented lobes + core & 1 4c 31.06 & @xmath108 & 0.70 & extended to 90 kpc ( presumably lobes ) & 2,3,4 3c 59 & @xmath109 & 0.72 & twin lobes + compact core & 5 pks 0235 + 023 & @xmath110 & 0.72 & twin lobes + compact core & 6 iras 0236.63101 & @xmath111 & & & 7 pks 034037 & @xmath108 & 0.87 & core with extension ( low - res .
map ) & 8 3c 93 & @xmath112 & 0.93 & twin lobes & 9 ms 0450.31817 & @xmath113 & & possible lobes + core & 10,11 pictor a & @xmath114 & 1.07 & edge - brightened double lobes & 1,12 b2 0742 + 31 & @xmath115 & 0.82 & twin lobes + compact core & 3,14 cbs 74 & @xmath116 & & core + possible lobes & 8,13,14 pks 085719 & @xmath117 & 0.80 & & 15 pks 0921213 & @xmath118 & 0.66 & twin lobes + compact core & 8,15,16 pks 1020103 & @xmath119 & 0.92 & faint lobes + compact core & 17 4c 36.18 & @xmath120 & 0.95 & twin lobes & 3,18 pks 115134 & @xmath121 & 0.76 & compact & 1,19 txs 1156 + 213 & @xmath122 & 0.78 & & 3 cso 643 & @xmath123 & 0.62 & asymmetric lobes + compact core & 8,14 3c 303 & @xmath124 & 0.81 & twin lobes + compact core & 20 3c 332 & @xmath125 & 0.92 & twin lobes + compact core & 5 arp 102b & @xmath126 & 0.22 & compact with extended halo & 21 pks 1739 + 18c & @xmath127 & 0.83 & twin lobes + faint core & 3,8,17 3c 382 & @xmath128 & 0.67 & twin lobes + compact core & 5 3c 390.3 & @xmath129 & 0.62 & twin lobes + compact core & 20 pks 191445 & @xmath130 & 0.94 & & 22 pks 230018 & @xmath131 & 0.77 & asymmetric lobes + compact core & 15,19 lllllll 3c 17 & @xmath132 & 0.12.4 & @xmath133 & @xmath134 & @xmath135 & 1 + 4c 31.06 & @xmath136 & 0.12.4 & @xmath137 & @xmath138 & @xmath139 & 1 + 3c 59 & @xmath140 & 0.12.4 & @xmath141 & @xmath142 & @xmath143 & 2 + pks 0235 + 023 & @xmath144 & 0.12.4 & @xmath145 & @xmath146 & @xmath147 & 2 + iras 0236.63101 & @xmath148 & 0.12.4 & @xmath149 & @xmath150 & @xmath151 & 3 + pks 034037 & @xmath152 & 0.12.4 & @xmath153 & @xmath154 & @xmath155 & 2 + 3c 93 & @xmath156 & 0.33.5 & @xmath157 & @xmath158 & @xmath159 & 4 + ms 0450.31817 & @xmath160 & 0.33.5 & @xmath161 & @xmath162 & @xmath163 & 5 + pictor a & @xmath164 & 0.12.4 & @xmath165 & @xmath166 & @xmath167 & 6 + b2 0742 + 31 & @xmath168 & 0.12.4 & @xmath169 & @xmath170 & @xmath171 & 2 + cbs 74 & @xmath172 & 0.12.4 & @xmath141 & @xmath173 & @xmath174 & 7 + pks 0921213 & @xmath175 & 0.12.4 & @xmath176 & @xmath134 & @xmath177 & 8 + pks 1020103 & @xmath140 & 0.12.4 & @xmath178 & @xmath179 & @xmath180 & 2 + 4c 36.18 & @xmath181 & 0.12.4 & @xmath137 & & & 1 + cso 643 & @xmath182 & 0.12.4 & @xmath183 & @xmath170 & @xmath184 & 2 + 3c 303 & @xmath185 & 0.53.0 & @xmath165 & @xmath186 & @xmath187 & 9 + 3c 332 & @xmath188 & 0.53.5 & @xmath189 & @xmath190 & @xmath177 & 10 + arp 102b & @xmath191 & 0.43.8 & @xmath192 & @xmath193 & @xmath194 & 11 + pks 1739 + 18c & @xmath195 & 0.12.4 & @xmath196 & & & 2 + 3c 382 & @xmath197 & 0.12.4 & @xmath198 & @xmath133 & @xmath199 & 2 + 3c 390.3 & @xmath200 & 0.53.0 & @xmath155 & @xmath201 & @xmath202 & 9 + pks 191445 & @xmath203 & 0.12.4 & @xmath155 & @xmath204 & @xmath205 & 12 + pks 230018 & @xmath206 & 0.12.4 & @xmath207 & & & 8 + | we report the completion of a survey of radio - loud active galactic nuclei ( agns ) begun in an earlier paper with the main goal of finding and studying broad , double - peaked balmer lines .
we present h spectra of 13 more broad - lined objects , including 3 with double - peaked h profiles .
the final sample includes 106 radio - loud agns . in our final census 20% of objects
have h lines with double peaks or twin shoulders ( the `` double - peaked emitters '' ) and of these , 60% ( the disk - like emitters ) can be fitted quite well with a model attributing the emission to a circular , relativistic , keplerian disk . in four objects where
broad h and lines have been observed , we compare the profiles with models of photoionized accretion disks and find them to be in reasonable agreement .
we reaffirm the conclusion of paper i that double - peaked emitters stand out among radio - loud agns on the basis of a number of additional properties that they possess : ( i ) an unusually large contribution of starlight to the optical continuum around h , ( ii ) unusually large equivalent widths of low - ionization lines ( [ ] and [ ] ) , ( iii ) unusually large [ ] / [ ] ratios , and ( iv ) balmer lines which are on average twice as broad as those of other radio - loud agns and preferentially redshifted .
we consider and evaluate models for the origin of the lines and we find accretion - disk emission to be the most successful one because it can explain the double - peaked line profiles and it also offers an interpretation of the additional spectroscopic properties of these objects .
we find the alternative suggestions ( binary broad - line regions , bipolar outflows , anisotropically illuminated spherical broad - line regions ) unsatisfactory because ( a ) they fail direct observational tests , ( b ) they can not explain all of the unusual properties of disk - like emitters self - consistently , or ( c ) in one case the physical foundations appear to be unsound .
we suggest that double - peaked emitters and accretion - powered liners are the segment of the agn population in which the accretion rate is considerably lower than the eddington rate , with the consequence that the inner accretion disk takes the form of an ion torus and the wind that normally enshrouds the disk proper is absent . |
a symbiotic star is a binary composed of a red giant and a hot blue component , often an accreting white dwarf .
many symbiotic stars have been detected as soft x - ray sources using _ rosat _ @xcite .
the _ rosat _
detections can usually be classified into two types .
one is of the supersoft type , which can be understood as photospheric emission from the hot white dwarf .
the second type is optically thin plasma emission with temperatures of order 10@xmath1k , which attributed to the colliding winds .
that is , a strong , fast ( @xmath21000 kms@xmath3 ) wind from the vicinity of the accreting white dwarf , which , in many cases , are known to be present through studies of uv and optical emission lines , is colliding with the dense , low - velocity wind from the m giant , shock - heating the gas to x - ray emitting temperatures .
more recently , a subset of symbiotic stars have been shown to be a strong source of hard ( @xmath410 kev ) x - rays in the _ integral _ galactic plane scan and in the _ swift _ bat survey ( @xcite and references therein ) , which requires the presence of yet another type of x - ray emission in these symbiotic stars .
ch cygni is an oft - observed , yet poorly - understood symbiotic star , at a distance of 245@xmath550 pc according to the _ hipparcos _ parallax measurements @xcite .
it has produced radio @xcite and optical @xcite jets , which appear to be ejected in the plane of the sky as seen from the earth , implying an accretion disk and/or binary with an inclination angle of nearly 90@xmath6 .
@xcite analyzed a large body of infrared spectra and proposed ch cyg to be a triple system , consisting of a 756 day period inner binary in an 14.5 yr orbit around a third component .
however , @xcite cautioned against the triple system interpretation on several grounds .
in particular , the orbital solution of requires a low inclination angle , contrary to the indications from the jet data .
in addition , such a low inclination then results in the unrealistically low mass ( 0.2m@xmath7 ) for the white dwarf . to add to the confusion
, @xcite claimed that ch cyg was a triple system as proposed by @xcite but both the inner and the outer binaries were eclipsing , thus requiring high inclination angles for both .
more recently , the similarity between the 756 day period of ch cyg with non - radial pulsations in quite a few semi - regular variables has become apparent ; therefore , the 756 day period of ch cyg is likely to be a non - radial pulsation mode rather than an orbital period ( @xcite and references therein ) .
ch cyg has been known to be a variable x - ray source for many years ; the early x - ray data from _
heao-1 _ to _ exosat _ are summarized by @xcite .
it was detected with _
rosat _ showing the colliding wind - type spectrum and an additional hard component @xcite .
however , it was the _ asca _ observation that revealed the dramatic two - component x - ray spectrum @xcite .
there is a relatively unabsorbed soft component , dominant below 2 kev , that can be modeled as two - temperature ( [email protected] , 0.7 kev ) optically thin thermal plasma emission . above 2 kev
, there is a separate , highly absorbed hotter component ( kt@xmath27 kev ) .
@xcite interpreted the latter as due to accretion onto a @xmath40.44 m@xmath7 white dwarf , and the former as either from the m giant or from the jets .
@xcite proposed an alternative interpretation of a single spectral component seen through an ionized wind of the m giant .
hetg observation was performed to decide between the two models , but ch cyg was found to be much fainter than during the _ asca _ observation , making the grating data of little use .
instead , @xcite analyzed the 0th order data and discovered a weak , spatially extended emission from the jet , in addition to the much brighter , spatially unresolved , component .
here we report on two _ suzaku _ observations of ch cyg performed in 2006 .
we have also re - analyzed the _
data and the _ chandra _ zeroth order data , and present the results in a uniform manner .
we observed ch cyg twice with the japanese - us x - ray astronomy satellite , _ suzaku _ @xcite .
the exposure times with the x - ray imaging spectrometer ( xis ) instrument @xcite was approximately 35 ksec on both occasions .
we have analyzed the data prepared using the version 0.7 pipeline @xcite .
this results in an energy scale accurate to @xmath50.2% at fe k@xmath0 and @xmath55 ev below 1 kev @xcite .
although ch cyg is definitely detected with the pin detector of the hard x - ray detector ( hxd ; @xcite ) instrument at approximately 7@xmath8 ergs@xmath9s@xmath3 in the 1225 kev band , this is sufficiently close to the current systematic uncertainties in the background estimation @xcite that we have chosen to concentrates on the xis data .
we have extracted the xis data from a 2.5 arcmin radius extraction region centered on the source , used the 2.55.0 arcmin annulus as the background , and generated response files taking into account the vignetting of the x - ray telescopes ( xrt ; @xcite ) and the time - variable contamination on the optical blocking filter of the xis .
we have combined the spectra taken with the three units ( xis0 , xis2 , and xis3 ) with frontside illuminated ( fi ) ccds , and analyze the combined spectrum simultaneously with that taken with xis1 with the backside illuminated ( bi ) ccd .
our source extraction region contains roughly 93% of the source flux , and this is taken into account by the response file . on the other hand , it does not account for the fact that the remaining 7% of the flux is in the background region , and this has been subtracted away ; we have therefore multiplied the fluxes derived from the spectral fits by 1.08 ( @xmath10 ) .
the log of _ suzaku _ observations can be found in table[tab : obslog ] , together with a summary of the _ asca _ ( see @xcite for details ) and _ chandra _ ( see @xcite for details ) data .
[ cols= " < , < , < , < , < , < " , ] ch cyg has shown a dramatic spectral variability over the last dozen years or so , while keeping the same basic two - component structures . at the same time , _ swift _ bat 1425 kev light curve of ch cyg also shows a decline , from @xmath11 ctss@xmath3 in the 1425 kev band during the first 3 months of 2005 to an undetectable level ( @xmath12 ctss@xmath3 ) by the end of 2005 @xcite .
we first consider the nature of the hard x - ray decline from the _ asca _ era to the _ suzaku _ era .
the extraordinary strength of the 6.4 kev , fluorescent fe k@xmath0 line shows that our direct line - of - sight is now blocked and we only observe hard x - ray photons scattered into our line of sight @xcite .
the low - energy cut - offs of the hard component did not vary much among the four observations shown in figure[fig : unf ] .
the best - fit n@xmath13 values were 1.4 , 4.1 , 3.2 , and 2.2 @xmath14 @xmath9 , respectively in our fits to the four observations .
such levels of absorption , which are admittedly uncertain due to the relatively simple model we have adopted , are insufficient to explain the observed equivalent widths of the 6.4 kev line in case ii of ( spherically symmetric absorber , direct and scattered x - rays both observed at earth ) , and requires case iii ( direct x - rays hidden ) . in contrast , the fe k@xmath0 complex was dominated by the thermal lines in the _ asca _ data , so in that observation our direct line of sight was open . the continuum flux declined by a factor of @xmath230 .
in contrast , the 6.4 kev line fluxes declined from 3.5@xmath15 photons@xmath9s@xmath3 to 1.4 and 1.0@xmath15 photons@xmath9s@xmath3 , respectively , in the two _ suzaku _ observations .
it is therefore likely that all of the direct x - rays , as well as the majority ( 6070% ) of the scattered x - rays , are blocked during the _ suzaku _ era . as for the central engine that produces the hard x - ray emission ,
we consider it likely that this is due to accretion onto the white dwarf , following @xcite .
in fact , the temperature ( kt@xmath210 kev ) and the luminosity ( 10@xmath16 ergss@xmath3 ) seen by _ asca _ are both similar to that seen in the old nova , v603 aql @xcite .
if this analogy is correct , then the x - rays are probably generated in the boundary layer between the accretion disk and a relatively massive white dwarf . between the m giant wind and the corona , wind , and/or jet from the accretion disk ,
there are more than enough candidates for the scattering medium .
however , as @xcite have calculated , typical line - of - sight column through the m giant wind in a symbiotic system is @xmath17 @xmath9 , which is borne out observationally by the widespread detection of the supersoft x - rays in symbiotic stars @xcite .
m giant winds are therefore far too optically thin to produce a significant amount scattered x - rays .
unless the white dwarf in ch cyg is unusually close to the photosphere of the m giant , which now seems unlikely with the preference for a 14.5 yr orbital period @xcite , or the m giant has much stronger mass loss than in other symbiotic systems , the m giant wind is therefore unlikely to be the site of hard x - ray absorption and scattering .
moreover , the strong decrease in the 6.4 kev flux also points to a relatively compact site for the scattering , i.e. , material associated with the accretion disk , such as an accretion disk corona or an accretion disk wind . what about the compton - thick material that is required to block out the direct hard x - rays ?
the m giant itself is a candidate .
however , according to @xcite , the eclipse of the 14.5 yr binary would have happened around 2000 .
thus , the existing orbital solution of ch cyg suggests that the 2006 low state is not due to an eclipse by the m giant .
we therefore consider it likely that the blockage was caused by the accretion disk .
recall that jet observations indicate a high inclination angle for the disk @xcite , very close to 90@xmath6 .
perhaps the disk thickness varies from epoch to epoch ; perhaps the disk is slightly tilted and precesses . in either case
, it seems reasonable for a highly inclined disk to be able to block our direct line - of - sight .
if the scattering medium is , say , a relatively compact accretion disk corona , it is plausible for a tilted and/or thickened outer disk also to block a significant fraction of the scattered x - rays , as we have inferred from the decline in fluorescent fe line flux .
if this picture is correct , the same thickening or precession of the accretion disk may also explain some of the optical / uv variability of ch cyg .
new optical / uv observations in the _ swift _ era will enable a direct comparison of hard x - ray and optical / uv variabilites . what can we learn about the soft component from the _ suzaku _ data ?
we believe we now have sufficient data to disprove the ionized absorber interpretation of @xcite .
first , the analysis of @xcite showed that while the hard component was significantly variable , the soft component was consistent with poisson noise .
this is difficult to explain in the ionized absorber model .
second , we do not believe that the compton - thick material blocking our line of sight to the hard x - ray emission region can be significantly ionized , given the modest x - ray luminosity ( of order 10@xmath16 ergss@xmath3 ) of ch cyg .
finally , the he - like line of mg is clearly seen in figure[fig : xis ] , while the h - like line is not . since the latter is stronger in any plasma hotter than [email protected] kev , we conclude that the soft component is dominated by plasma cooler than [email protected] kev .
this is also consistent with the strengths of other emission lines ( o , fe l , ne ) seen below 1 kev , which would be significantly diluted if much hotter plasma co - existed in the emission region .
we therefore conclude that the soft x - rays indeed are a distinct component .
the emitting region must be separate from , or significantly more extended than , the hard x - ray emitting region .
on the other hand , the bulk of the soft x - rays were spatially unresolved with _
chandra _ @xcite .
this sets the maximum size to be of order 100 au ( 0.4@xmath18 at 250 pc ) , or 800 light minute in radius .
the entire binary is likely to be smaller than this , assuming a 14.5 yr orbital period .
an extended emission region , comparable to the size of the binary , is expected in the colliding wind interpretation , in which the m giant wind is colliding with an outflow from the vicinity of the accreting white dwarf .
the exact nature of the latter is unclear ; it can be a compact jet ( unlike that produced the spatially - resolved x - rays observed with _
chandra _ ; @xcite ) , or it can be a less collimated outflow , more appropriately termed wind .
colliding wind systems are usually not highly variable on short time scales , presumably because any local instabilities are averaged over the large emission site . only a global change ( changes in binary separation or in the wind mass loss rate )
can modulate the x - ray flux significantly .
if the colliding wind emission is from a 10 au size region , and the accretion disk wind velocity is 1,000 kms@xmath3 , then we do not expect significant modulation in soft x - rays on time scales shorter than about 2 weeks from a `` wind travel time '' argument .
this consistent with the lack of soft x - ray variability seen with _
asca _ @xcite .
@xcite considered the soft component to be from the corona of the m giant .
however , m giants are in the `` coronal graveyard '' to the upper right of the dividing line in the hr diagram .
_ chandra _ observations of single giants in this region show that coronal emissions are far too low a level to explain the soft component of ch cyg : arcturus is detected at 3@xmath19 level at a 0.22 kev luminosity of @xmath20 ergss@xmath3 and the upper limit for aldebaran is 7@xmath21 @xcite ; the m supergiant betelgeuse has a _
upper limit of @xmath22 ergss@xmath3 for coronal temperature @xmath23 k @xcite . whenever m giants are detected in x - rays , they are usually interpreted as due to a binary companion ( @xcite and references therein ) .
although binarity can affect the x - ray luminosity , this is limited to short period systems ( periods of 2 days 2 weeks for rs cvn systems , for example ) where tidal torque would lead to synchronous rotation , which is unlikely in a symbiotic binary .
for the coronal model of the soft x - ray emission to succeed , the mass donor in the ch cyg system must be a truly exceptional m giant .
so far , we have concentrated on the comparison between _ asca _ and _ suzaku _ data .
what about the _ chandra _ observation , during which both the soft and the hard x - ray components were fainter and there was no detection of the fluorescent fe k@xmath0 line ? for this , we must invoke a separate mechanism .
if the accretion rate onto the white dwarf declines dramatically , we expect both the hard and the soft x - ray luminosities to drop , and this appears to be what happened in 2001 .
our two _ suzaku _ observations of ch cyg have caught the system in a soft x - ray bright , hard x - ray dim state . based on the fe k@xmath0 spectra , we believe we currently observe only the scattered hard x - ray component . the decline in the hard x - ray flux is only an apparent decline , and the intrinsic x - ray luminosity of ch cyg may still be as high as during the _ asca _ observation , though the actual value is unknowable .
a symbiotic star such as ch cyg is filled with possible scattering media , but we favor material above the accretion disk as the scattering site . as for the object that blocks our direct view of the hard x - ray emission , we believe it is more likely to be an edge - on accretion disk than the m giant mass donor .
the hard x - rays presumably originate from accretion onto the white dwarf , but further progress would require high - quality x - ray spectroscopy when the direct view is not blocked .
ayres , t.r . , brown , a. , harper , g.m .
2003 , , 598 , 610 ezuka , h. , ishida , m. , makino , f. 1997 , , 499 , 388 galloway , d.k . ,
sokoloski , j.l .
2004 , , 613 , l61 hinkle , k.h . ,
fekel , f.c . ,
johnson , d.s . ,
scharlach , w.w.g .
1993 , , 105 , 1074 inoue , h. 1985 , sp.sci.rev , 40 , 317 kennea , j.a . ,
mukai , k. , .... 2006 , in preparation kokubun , m. et al .
2006 , , in press koyama , k. et al .
2006 , , in press leahy , d.a . ,
taylor , a.r .
1987 , , 176 , 262 mitsuda , k. et al .
2006 , , in press mukai , k. , orio , m. 2005 , , 622 , 602 munari , u. , yudin , b.f . ,
kolotilov , e.a . ,
tomov , t.v .
1996 , , 311 , 484 mrset , u. , wolff , b. , jordan , s. 1997 , , 319 , 201 perryman , m.a.c .
, et al .
1997 , , 176 , 262 posson - brown , j. , kashyap , v.l . ,
pease , d.o .
, drake , j.j .
2006 , , submitted ( astro - ph/0606387 ) schmidt , m.r . ,
zacs , l. , mikolajewska , j. , hinkle , k.h .
2006 , , 446 , 603 serlemitsos , p.j .
2006 , , submitted .
skopal , a. , bode , m.f . ,
lloyd , h.m . , tamura , s. 1996 , , 308 , l9 solf , j. 1987 , , 180 , 207 takahashi , t. et al .
2006 , , submitted .
taylor , a.r . ,
seaquist , e.r . ,
mattei , j.a .
1986 , , 319 , 38 van den berg , m. , grindlay , j. , laycock , s. , hong , j. , zhao , p. , koenig , x. , schlegel , e.m . , cohn , h. , lugger , p. , rich , r.m . ,
dupree , a.k . ,
smith , g.h . ,
strader , j. 2006 , , 647 , l135 wheatley , p.j .
2001 , aip conf .
599 , x - ray astronomy : stellar endpoints , agn , and the diffuse x - ray background ( melville : aip ) , 1007 | ch cygni is a symbiotic star consisting of an m giant and an accreting white dwarf , which is known to be a highly variable x - ray source with a complex , two - component , spectra .
here we report on two _ suzaku _ observations of ch cyg , taken in 2006 january and may , during which the system was seen to be in a soft x - ray bright , hard x - ray faint state . based on the extraordinary strength of the 6.4 kev fluorescent fe k@xmath0 line , we show that the hard x - rays observed with _ suzaku _ are dominated by scattering . |
it has long been suggested that measuring the density and temperature distributions of the intergalactic gas would allow a cluster mass determination free from the limitations inherent in the virial estimates based on galactic velocities ( bahcall & sarazin 1977 ; mathews 1978 ) .
the necessary requirement that the gas be in hydrostatic equilibrium in the cluster gravitational well is largely met in most regular clusters ( e.g. , sarazin 1988 ; navarro , frenk , & white 1995 ; this issue is discussed in more detail in 4.2 ) . the gas density profile for symmetric clusters
can readily be obtained with an imaging instrument , such as or .
obtaining temperature distributions has proven to be more problematic . in the absence of spatially resolved spectral data ,
much work has been done assuming constant temperature . at small cluster radii and outside of the central cooling flow regions ,
this assumption indeed applies ( e.g. , watt et al.1992 ; ikebe et al . 1996 ; loewenstein & mushotzky 1996a )
. however , mass estimates at large radii are only as good as the temperature measurements at those radii .
this is illustrated by large uncertainties of the inferred mass when allowance for nonisothermality is made ( e.g. , henry , briel , & nulsen 1993 , hereafter hbn , obtain a 68% range spanning a factor of three for the a2256 mass within @xmath7 ; see also loewenstein 1994 ) .
cosmological simulations predict that cluster gas temperatures decline at large radii ( e.g. , tsai , katz , & bertschinger 1994 ; navarro et al .
1995 ; evrard , metzler , & navarro 1996 ) . at the time of this writing , however , only coma and a2163 have temperature measurements at radial distances exceeding @xmath8 mpc .
the x - ray satellite ( tanaka , inoue , & holt 1994 ) is capable of measuring the cluster temperature spatial distributions .
it indeed observes a significant temperature decline at large off - center distances in a number of hot , massive systems , such as a2163 ( markevitch et al .
1996 , hereafter m96a ) , a2256 , a2319 , a665 ( markevitch 1996 , hereafter m96 ) ; a2218 ( loewenstein 1997 ) ; and probably , in the cooler a3558 ( markevitch & vikhlinin 1997 , hereafter mv ) and hydra - a ( ikebe et al .
earlier and data on coma ( hughes , gorenstein , & fabricant 1988 ; hughes 1991 ) and results on perseus ( eyles et al .
1991 ) suggested qualitatively similar temperature behavior .
henriksen & white ( 1996 ) infer a radial temperature decline for several clusters from comparison of spectra obtained with and sss with different fields of view .
if declining temperature profiles are typical ( as is indeed suggested by our work in progress ) , it would imply that there is in fact less dark matter in cluster outskirts and probably more in the inner parts compared to the results derived under the isothermal assumption . in this paper
, we will quantify this effect for a2256 and discuss some of its implications .
table 1 parameters of image components from pspc [ cols="<,^,^,^,^,^ " , ] @xmath9 coordinates are fixed at the optical position of the brightest galaxy ; gaussian brightness profile width is fixed at @xmath10 .
a2256 ( @xmath11 ) is among those clusters with a detected temperature decline .
although it may not be the best candidate for a convincing mass measurement due to the presence of substructure ( discussed in 4.1 ) , the quality of the available x - ray data for this cluster is unsurpassed .
there are six deep pspc pointings covering different cluster regions out to @xmath12 mpc ( briel et al .
1991 , hereafter b91 ) and a high statistical quality temperature map for @xmath13 mpc with useful spatial resolution ( m96 ) .
a2256 is therefore a good starting point for a mass determination using a measured temperature profile .
a similar earlier effort on the more distant a2163 ( m96a ) has suggested that the dark matter density in that cluster falls off rather steeply with radius but also suggested that a2163 may be out of hydrostatic equilibrium at the radii of interest .
more evidence has since been added by squires et al . (
1997 ) that a2163 indeed is an advanced merger .
below , essentially the same method as that used in m96a is applied to a2256 , which is better resolved and more likely to be in hydrostatic equilibrium .
we parameterize @xmath14 ; all quoted errors are 90% one - parameter intervals .
b91 have shown that a2256 is a double x - ray cluster .
in addition , one of the x - ray brightness peaks coincides with the brightest cluster galaxy and is displaced by about 1.7 from the primary cluster s centroid . to derive the density profile of the primary cluster ,
we have fitted the pspc image by two symmetric projected @xmath15-models ( e.g. , jones & forman 1984 ) representing the two subclusters , their @xmath16 , @xmath15 , center positions and relative normalizations and the x - ray background being free parameters , and a gaussian brightness peak at the position of the central galaxy ( fixing @xmath10 as suggested by the hri image ) .
pspc data are analyzed as described in mv . for image analysis , only energies 0.72 kev were used to optimize statistics , and the psf model was taken from hasinger et al .
the best - fit centroid coordinates , luminosities and gas distribution parameters for the components of the fit are summarized in table 1 . for the primary cluster ,
we obtain @xmath17 mpc and @xmath18 for the emission measure profile ( confidence contours are shown in fig . 1 ) , taking into account the small effect of the observed temperature decline ( m96 and 3 ) and excluding the detectable point sources .
these parameters are derived using the data from @xmath19 for self - consistent use in the analysis below . the corresponding central electron density ( excluding the central galaxy )
is @xmath20 @xmath21 .
our simplifying assumption that the central galaxy is projected rather than embedded in the primary cluster , introduces a small inaccuracy of the central density distribution of the main cluster .
however , this inaccuracy is insignificant comparing to the uncertainty in the central temperature which will be obtained below ; we discuss this issue further in 5.1 . regarding the adequacy of the @xmath15-model at large radii
, the primary cluster emission can still be traced at the @xmath22 level , including the background uncertainty , out to @xmath23 , which is twice the virial radius ( the radius inside which the cluster mean density is 180 times the closure density ; @xmath24 mpc for a2256 using our mass measurement ) . using data within @xmath23
, we obtain @xmath25 and @xmath26 , not significantly different from the above values .
this is contrary to the simulation predictions of steepening gas profiles with radius ( e.g. , navarro et al .
1995 ; see , however , frenk et al .
fixing @xmath16 and using only the profile in the several smaller intervals of radii , we find that the global value of @xmath15 represents the local density slope to about @xmath27% accuracy in the inner cluster part . at the radii where the background uncertainty dominates ,
the possible error is greater : for @xmath28 , we obtain @xmath29 .
these uncertainties will be included in the mass estimates discussed below . for the smaller subcluster , @xmath30 and @xmath31 .
these parameters are poorly constrained but are not important for our final results , thus in the following analysis they are fixed at their best - fit values .
the two subclusters 0.52 kev luminosities within @xmath32 mpc are given in table 1 .
the smaller subcluster is two times less luminous than the primary cluster ; this differs from the finding of b91 ( 1/5 of the primary cluster s luminosity ) .
the difference is due to the fact that b91 refer to an excess flux in the 90 image sector containing the subcluster , while we refer to the full flux in the subcluster s @xmath15-model .
fitting the pspc spectrum of the subcluster while subtracting the emission of the primary component similarly to b91 , fixing the value of @xmath33 and using the 0.52 kev band ( as in mv ) , we obtain @xmath34 ( 2.58.2 ) kev , consistently with the result given below .
the pspc - measured profile of projected temperatures in the inner @xmath35 mpc of a2256 agrees remarkably well with the radial temperature gradient ( mv ) , but due to its relatively large errors is not used in the mass analysis below .
2 shows pspc temperature measurements at still greater radii , @xmath36 mpc ( excluding the 90sector with the smaller subcluster ) , obtained using the same procedure as in mv .
although poorly constrained ( mostly due to the background uncertainty ) , they suggest that the gas temperature continues to fall with radius .
while keeping this in mind , we will not use these data below .
more important for our analysis is the fact that the gas emission is still detected with 90% confidence at @xmath37 mpc .
we will conservatively interpret this as a lower limit on the gas temperature , @xmath38 at that large radius , and apply this limit to the hydrostatic temperature profiles in 5 .
( 0,8.0)(18.8,20 ) ( -0.2,20.1)=8.9 cm ( 9.5,20.1)=8.9 cm ( -0.1,11.2 ) = 3.5 mm fig .
1.two-parameter 90% confidence regions on @xmath16 and @xmath15 of the primary cluster gas density profile from and . for
, all temperatures and the spectrum of the central source were free parameters . for both instruments , parameters @xmath16 and @xmath15 of the secondary cluster were fixed at their best - fit values ; fit excluded the region of the central source .
( 9.6,11.2 ) = 3.5 mm fig .
2.temperature profile of the primary cluster .
values correspond to step - like profiles in projected annuli or spherical shells ( upper axis gives conversion to linear distances ) , while values are for projected annuli .
the central low temperature shows an additional component required in the central region .
errors are 90% . the lower limit at @xmath39 reflects the detection of the cluster emission there .
smooth lines show a representative set of models ( not projected and not showing the cooler component ) allowed at the 99% confidence by the data and the 40 limit ( see text ) .
technical details of the analysis of a2256 , including discussion of the systematics , are given in m96 , who presented a two - dimensional map of projected gas temperatures in this cluster .
the analysis technique used there and in this work , which efficiently takes into account the complex psf ( takahashi et al .
1995 ) , is described in m96a .
the cluster temperature map shows that except for a cooler subcluster , there is no significant asymmetry in the radial decline of the projected temperature .
here we reconstruct a three - dimensional , as well as projected , radial temperature profile of the primary cluster . in this section
, we make no assumptions regarding the underlying mass distribution to indicate the intrinsic data constraints , while in 5 , continuous hydrostatic profiles will be constrained by the data . as in 2
, we assume that the main cluster is spherically symmetric and that the smaller subcluster and the central source are projected onto it .
it is also assumed for simplicity that the subcluster is isothermal , since it only contributes about 20% of the flux in the image region that determines the temperature in our most important outer spherical bin ( for the same reason , the uncertainty in its brightness distribution is unimportant ) . under these assumptions ,
we first fit uniform temperatures in four spherical shells of the primary cluster with radii corresponding at the cluster s distance to @xmath40 , @xmath41 , @xmath42 and @xmath43 . together with them
, we also fit a temperature of the subcluster and various models for the central source , allowing its normalization to be a free parameter .
the image regions from which spectra are collected are annuli defined by the above radii excluding the 90 sector containing most of the subcluster , this sector , and an @xmath44 circle around the central galaxy .
gis and sis spectra from all regions are fitted simultaneously . due to the limited angular resolution , it is difficult to constrain both the temperature and spatial brightness distributions with alone .
we therefore take advantage of the information on the relative brightness distribution . because the resulting temperatures are sensitive to the assumed brightness distribution
, one must ensure consistency between the two instruments on this issue and that the use of the image does not introduce an error in the temperature measurement .
for this , we fix @xmath16 and @xmath15 of the subcluster and the positions of the three model components at their values , but free the temperatures , relative normalizations of the model components ( that is , the main cluster , the subcluster , and the central source ) , and @xmath16 and @xmath15 of the primary cluster , thus using almost no information in the fit .
confidence contours for the latter two parameters are shown in fig . 1 . and
-derived brightness distributions are in perfect agreement , while the accuracy is superior . therefore , @xmath16 and @xmath15 can be fixed at their values .
assuming that the central source has the same temperature as the surrounding gas yields a gas temperature of @xmath45 kev in the central @xmath46 spherical bin .
this may misleadingly suggest that the cluster central part is largely isothermal .
however , if that source is allowed to have a different spectrum , @xmath47 is significantly reduced by 1318 for one additional parameter depending on the assumed model , for which we tried a second thermal component , a cooling flow or a power law .
a power law model ( with the best - fit photon index @xmath48 ) is formally the best fit , and we chose to use this model in further analysis ( being aware that the real spectrum is probably still more complex ) . all models result in allowing a higher ambient gas temperature .
in fact , due to the limited angular resolution , we can not precisely localize the source of this additional emission : it may , for example , be one or more of the several radio sources in the central region ( rttgering et al .
there are no bright point sources in the relevant region of the hri image , which argues in favor of either thermal or nonthermal extended component .
however , its exact origin is not important for our analysis , therefore we assume it is the central galaxy .
this spectral complexity , combined with projection , results in a relatively large uncertainty in the central temperature of the main gas component , which will complicate an accurate measurement of the mass distribution in the central region . the best - fit temperatures in the radial shells of the primary cluster with other parameters treated as described above ,
are shown in fig . 2 as thick crosses .
for this model , @xmath49 for 685 d.o.f .
, reflecting our conservatism in accounting for the instrumental uncertainties .
for a consistency check and to show the intrinsic data constraints separated from the uncertainty of the deprojection , we performed a similar analysis assuming uniform temperatures within projected regions on the sky rather than in spherical shells ( even though it has less physical meaning ) .
these temperatures are also shown in fig . 2 as thin solid crosses .
the projected and three - dimensional temperatures differ as expected , with uncertainties of the latter being greater due to the deprojection .
note that the step - like profiles assumed above have little physical meaning ( since the real distribution is continuous ) , which , together with the natural anticorrelation of the errors in the adjacent spherical bins of the deprojected profile , explains differences such as that in the third annulus in fig .
this is not important for our mass analysis in 5.1 , in which continuous temperature profiles will be compared directly to the data . for the smaller subcluster we obtain @xmath50 kev and a relative iron abundance @xmath51 , exceeding with marginal significance the value for the primary cluster , @xmath52 .
note that the temperature and the luminosity of the subcluster agree very well with the @xmath53 relation of david et al .
the iron abundance in the outermost parts of the primary cluster is unconstrained ; we assume it uniform and fix at the best - fit value .
before using x - ray data to derive the distribution of dark matter in a2256 under the assumption of hydrostatic equilibrium , we should address the adequacy of this assumption .
fabricant et al . (
1989 ) , b91 , and davis & mushotzky ( 1993 ) showed that a2256 contains substructure both in the optical and x - ray data .
this cluster has a radio halo and head - tail radio sources ( rttgering et al .
1994 ) which may be related to a merger ( although their relation is not unambiguously established ) .
a merger would result in significant gas turbulence and inadequacy of our assumption of spherical symmetry .
however , there is a strong argument against an advanced merger , which is the absence of any significant asymmetries of the gas temperature ( m96 ) expected in a collision of such massive clusters ( e.g. , schindler & mller 1993 ) .
the detection of hot spots by briel & henry ( 1994 ) was later not confirmed by reanalysis of their data using an updated pspc calibration ( mv ) .
the accuracy of the measurement in m96 was adequate for detecting a temperature pattern characteristic of a merger , and indeed such a pattern was observed in several clusters , including the obviously merging a754 ( henriksen & markevitch 1996 ) .
the observed probable difference in iron abundances of the subclusters ( 3 ) also argues against a late - stage merger since the gas would mix if clusters had passed through each other .
another argument is the presence of a remarkably long and straight narrow tail radio source near the cluster center ( rttgering et al .
its existence strongly suggests that there is no significant gas turbulence there .
we therefore assume that the merger has not proceeded far enough to disturb the bulk of the primary cluster s gas . moreover , below it will be assumed for simplicity that the two clusters are physically well separated along the line of sight , so that the gravitational effect of the subcluster on the primary cluster s gas can be neglected .
indeed , such a separation is suggested by the galaxy velocity data .
fabricant et al.(1989 ) and b91 noted that the cluster line - of - sight velocity distribution can be described by two components , correspondent to the two subunits , separated by about 2000 . a simple calculation ( as in , e.g. , henriksen & jones 1996 ) shows that , assuming two clusters of comparable masses falling from rest at some large distance toward each other , a mass of the main cluster of @xmath54 as obtained below , and the observed projected distance between the cluster centers , this rather small relative velocity implies a separation of @xmath55 mpc . in a similar analysis of a2163 , m96a raised the possibility that plasma in the outer , low density region may be out of electron - ion temperature equipartition , in such a way that the measured @xmath56 underestimates the thermodynamic temperature . in a2256 ,
the temperature measurements correspond to smaller radii and higher densities . in the radial bin
where detects a temperature drop , the timescale for reaching equipartition via collisions is @xmath57 yr , shorter than the time which has to have passed since the last merger for the cluster to become relaxed as we observe it .
actual temperature measurements beyond this radius will not be used ( only a lower limit , @xmath38 ) .
the temperature decline in a2256 is also less steep compared to that in a2163 where it implied convective instability . at the radii of the temperature decline in a2256 ,
the temperatures correspond to a polytropic index of 1.50 ( 1.241.9 ) .
therefore , the gas should be convectively stable on large scales .
there is a possibility that some of the gas in the outer cluster is confined by the local potential wells of small infalling subunits , rather than that of the primary cluster .
if that is the case one would not expect the cluster to be azimuthally symmetric , while the image shows it is rather smooth out to large radii , except for a small ellipticity . at large cluster radii
, residual bulk motions of gas from past merger events and infall may become non - negligible , which would mean that hydrostatic equilibrium may not be an adequate assumption .
simulations of evrard et al.(1996 ) showed that the kinetic energy of these motions is between 070% of the gas thermal energy at the cluster virial radius , with high values corresponding to clusters experiencing mergers .
given that the bulk of a2256 seems to have been undisturbed in the recent past as we concluded above , we can assume that a2256 is relatively well virialized within our radius of interest .
a significance of this effect , as well as the uncertainty arising from our spherical symmetry assumption , can best be assessed through the hydrodynamic simulations .
several independent authors ( e.g. , tsai et al.1994 ; schindler 1996 ; evrard et al . 1996 ;
roettiger et al . 1996 ) applied the x - ray mass estimation algorithm , which assumes spherical symmetry and hydrostatic equilibrium , to the simulated clusters in various cosmological backgrounds .
they find that , when the actual temperature profile is used ( as we do ) , such estimates are on average unbiased and have an _ rms _ scatter around the true mass of about 15% at the core and about 30% at the virial radius .
these are conservative estimates of the scatter because clusters with obvious mergers , which show the biggest mass errors , are included in them but would be excluded by the observers ( recall that the primary cluster of a2256 is not a merger , as we concluded above ) .
bartelmann & steinmetz ( 1996 ) find their simulated x - ray mass estimates to be systematically biased , and attribute the discrepancy with other authors to their assumption of more severe observational conditions , which are not relevant to our present work .
the values above may therefore be regarded as conservative estimates of the systematic uncertainty of an individual cluster mass determined by the x - ray method .
we will continue this discussion in 6.1 in application to our results .
the x - ray data will now be used to constrain the mass profile of the primary cluster .
because the gas temperature profile is known much less accurately than the gas density profile , we chose to use the fitting method of hughes ( 1989 ) and hbn with some technical variations .
two functional forms for the dark matter radial profile will be considered , which together can approximate just about any physically meaningful symmetric distribution : one with a constant core , @xmath58 and one with a central singularity ( cusp ) predicted by cluster simulations ( e.g. , navarro et al .
1995 ) , @xmath59 in both profiles , the density in the outer part declines asymptotically as @xmath60 .
navarro , frenk , & white ( 1997 ) found that simulations in different cosmologies produce a `` universal '' density profile for clusters in equilibrium ; it is of the latter form with @xmath61 and @xmath62 .
the isothermal sphere model corresponds to @xmath63 ; profiles with @xmath64 are expected under different assumptions about the halo formation ( e.g. , syer & white 1997 and references therein ; kofman et al.1996 ) . for the gas ,
we use a @xmath15-model density profile ( cavaliere & fusco - femiano 1976 ) , @xmath65 with @xmath16 and @xmath15 derived in 2 . for simplicity ,
the relatively small ( of the order of 1% , hbn ) contribution of galaxies to the total mass is neglected . with the above functional forms for the major mass components , we solve the hydrostatic equilibrium equation ( e.g. , sarazin 1988 )
analytically ( see appendix ) to recover the radial gas temperature profile for a given set of parameters @xmath66 , @xmath67 ( if applicable ) , @xmath68 , the dark matter normalization @xmath69 , and the temperature at zero radius @xmath70 .
the general behavior of the solutions is discussed in detail by e.g. , hughes ( 1989 ) , hbn , and loewenstein ( 1994 ) . for dark matter profiles of the cusp form ( 2 ) and a @xmath15-model gas density profile ,
the gas temperature profile at @xmath71 is finite if @xmath72 and has zero derivative if @xmath73 . in practice ,
the insufficient spatial resolution of the temperature data does not allow us to independently constrain all three shape parameters of the profile ( 2 ) , thus we chose to fix @xmath61 as suggested by simulations .
temperature profiles rising to infinity outside the measurable region are not allowed in our analysis ( which in practice does not affect our constraints ) , but we do allow them to fall to zero at a finite radius ( outside @xmath39 where certainly detects cluster emission ) , since hydrostatic equilibrium is not expected to hold far from the virial radius .
in this work , we are interested in the shape of the dark matter density distribution .
therefore , for a given set of the shape parameters @xmath66 and @xmath68 , we fit @xmath69 and @xmath70 so that after the projection and convolution with the telescope responses , the temperature profile minimizes s @xmath47 .
the spectra and the normalizations of the additional central galaxy source and the subcluster are treated as free parameters to allow maximum freedom for the temperature profile of the main cluster .
the gas associated with the central galaxy is not included in the hydrostatic equilibrium equation , although in reality , the central galaxy atmosphere resides in the same gravitational potential and is in pressure equilibrium with the main cluster gas that is being modeled . in principle , its temperature can be modeled in a similar way and used to provide additional constraints on the mass distribution in the center .
however , in the absence of better - resolution data on whether the two gas phases cohabit in the same volume or one of them replaces the other , we chose a conservative treatment that does not additionally restrict the range of mass models and consists of completely freeing the normalization and the spectrum of the central galaxy component .
this approach should not significantly affect the mass measurements outside the @xmath74 central galaxy region .
the model is fit directly to the data ( not to the step - like temperature profiles shown in fig .
minimum @xmath47 values for both functional forms of the dark matter profile are very similar .
they are also similar to the value obtained for the step - like temperature distribution , thus this value can be considered an `` absolute '' minimum of @xmath47 with which to compare the models of different form . those models for which @xmath75 are rejected with 90% confidence . a profile acceptable for
is then checked against the requirement that the temperature be greater than zero at @xmath39 ( a lower limit from the flux detection ) and if not complying , rejected at the 90% confidence .
the actual temperature estimates between @xmath76 do not constrain profiles any better than this lower limit , thus for simplicity they are not used . however , these estimates reinforce the lower limit even if , contrary to our assumption , hydrostatic equilibrium does not hold at these large radii .
( 0,1.2)(18.8,20 ) ( -0.2,20.1)=8.4 cm ( 9.1,20.1)=8.4 cm ( 0.15,11.5)=8.4 cm ( 9.25,11.5)=8.4 cm ( 8.3,19.25)*_a _ * ( 17.5,19.25)*_b _ * ( 8.3,10.65)*_c _ * ( 17.5,10.65)*_d _ * ( -0.1,2.8 ) = 3.5 mm fig .
3.models whose temperature profiles are shown in fig . 2 : ( _ a _ ) their dark matter density profiles ; ( _ b _ ) corresponding enclosed total masses ( including gas ) ; ( _ c _ ) fractions of the gas density in the total local density at a given radius ; ( _ d _ ) fractions of gas in the total mass within a given radius .
values from the isothermal analysis ( for the cluster average @xmath77 kev ) are shown for reference .
acceptable profiles of both functional forms ( with a core or a central @xmath78 cusp ) are shown for @xmath79 .
absolute values are for @xmath80 ( total mass scales as @xmath81 and gas fraction as @xmath82 ) . a narrow range of hydrostatic temperature profiles is allowed by both and constraints .
representative examples of the temperature profiles corresponding to the models acceptable within the approximate combined 99% confidence region as described above are shown in fig . 2 ( note that the temperatures in the central @xmath83 may have little physical meaning due to our incomplete treatment of the central galaxy atmosphere , as discussed above ) .
the corresponding dark matter density and total mass profiles are shown in fig .
shallow mass distributions best describe the temperatures alone , but they predict a temperature drop to zero immediately beyond the field , contrary to the data .
the combined data exclude , at greater than 99% confidence , dark matter profiles with a core of the form ( 1 ) for which @xmath84 , and those of the form ( 2 ) for which @xmath85 , for any value of the scale parameter @xmath68 ( these constraints on @xmath66 in fact correspond to similar non - parametric slopes of the profile over the measured region ) .
very steep dark matter profiles with @xmath86 , which correspond to the outer cluster regions being dominated by gas , are not excluded by the data , although they lack physical motivation .
naturally , acceptable profiles with shallower outer slopes have smaller @xmath68 . for dark matter profiles with cores , @xmath87 ( that is ,
dark matter is more centrally peaked than gas ) , and for profiles with cusps , @xmath88 for the values of @xmath79 we considered . for a particular value of @xmath89 , the most acceptable outer slopes are @xmath90 and 3 for the core and the cusp profile forms , respectively ; thus the `` universal '' profile of navarro et al .
( 1997 ) appears to be acceptable . at @xmath91 mpc
( about half the virial radius ) where direct accurate temperature measurements exist , the total mass for the allowed models is @xmath92 ( for the mass values hereafter , we include in quadrature the uncertainties of the local gas density slope at the respective radii , as estimated in 2 ) .
this mass is consistent with the value of @xmath93 obtained earlier by hbn at this radius ( correcting for their distance error ) . at the radius of the mean overdensity 500 , which for a2256 corresponds to @xmath94 mpc using our mass profile ,
the mass is @xmath95 , significantly below the value of @xmath96 predicted by the scaling law of evrard et al .
this difference is due to the fact that the simulations typically produce a shallower temperature fall than the observed one .
if the cluster had been isothermal , its mass within @xmath97 would be @xmath98 , in agreement with the scaling relation ( in part because @xmath97 itself would be slightly larger ) . at @xmath99 mpc ( the virial radius )
, we obtain an unprecedentedly accurate estimate of the mass , @xmath100 . this value is significantly smaller , by a factor of 1.6 , than the value obtained assuming isothermal gas , as is naturally expected for a declining temperature profile ( e.g. , evrard et al .
the local dark matter density is constrained less accurately , but it is possible to derive a strong upper limit beyond @xmath101 mpc .
this limit is 1.52 times lower than the isothermal value , although it is of course rather model - dependent . because the central cluster temperature is poorly constrained ,
so is the density profile within the cluster x - ray core the profiles with and without a central density cusp are both acceptable .
nevertheless , the total mass internal to the core ( @xmath5 mpc ) is well constrained as @xmath102 for the acceptable models .
this is consistent with the _ bbxrt _ measurement ( restricted to @xmath103 mpc ; miyaji et al .
the formal confidence intervals for the values given above represent only the experimental uncertainties ( and as such are directly comparable to those from earlier works ) . because of
the high quality of the data , additional uncertainties of the method due to possible deviations from spherical symmetry and hydrostatic equilibrium , as predicted by simulations ( 4.2 ) , become comparable to or even exceed our experimental errors .
we do not attempt to quantify these systematic uncertainties for a2256 , since the main purpose of this paper is to compare the mass estimates obtained using the actual temperature profile with those from the isothermal analysis ( for which the same systematic deviations apply ) . to draw reliable quantitative conclusions about the cluster mass profiles in general ( which is our more distant goal ) ,
it is obviously necessary to study a sample of clusters , of which a2256 is the first member .
nevertheless , even for this individual cluster , the qualitative effects discussed below exceed the realistic uncertainties suggested by the simulations .
it is also noteworthy that the formal measurement accuracy is approaching the limit due to the unknown cluster peculiar velocities and hence accurate distances , @xmath104 .
the two components of a2256 have velocities differing by about 2000and thus their distances calculated from redshifts may be in error by as much as 57% , implying a similar error in mass within a given projected radius . in the absolute sense , this uncertainty is small and , of course , @xmath105 is not yet known to such accuracy .
however , relative results may be significantly affected .
for example , the observed difference of baryon fractions ( which are proportional to @xmath106 ) between the low-@xmath107 clusters a1060 and awm7 ( loewenstein & mushotzky 1996a ) might be entirely explained if either of them had a peculiar velocity of 8001000 .
the mass value we obtain within @xmath7 corresponds to a factor of 1.6 higher average baryon fraction ( which is higher or equal to the gas fraction , @xmath108 ) than the value obtained from the isothermal analysis .
this change , if applicable to other clusters , would make the observed cluster baryon fraction even more contradictory with the hypothesis of @xmath109 , as discussed e.g. , by white et al .
( 1993 ) , white & fabian ( 1995 ) , and most recently by evrard ( 1997 ) .
the gas fraction above formally implies that @xmath110 for @xmath80 if one assumes that the cluster matter content is representative of the universe as a whole , that dark matter is nonbaryonic , and the standard nucleosynthesis ( walker et al .
1991 ) . alternatively ,
assuming the high or low measured deuterium abundance ( see review in steigman 1996 ) it implies @xmath111 and 0.20.5 , respectively .
as figures 3(_c , d _ ) show , the gas fraction increases significantly with radius , by about a factor of 2.53 from the radius of overdensity @xmath112 ( @xmath113 ) to @xmath7 even if one considers only the least steep models .
similar behavior is obtained for a sample of other clusters under the assumption of gas isothermality and for several cool systems taking into account the observed temperature gradients ( david , jones , & forman 1995 ) .
note that a2256 differs from clusters in that sample by its steeper gas density profile and a considerably lower _ isothermal _ estimate of the gas fraction ( which is also almost uniform with radius ) .
however , the use of the actual temperature profile makes the value and the radial dependence of the gas fraction in a2256 rather similar to that for the rich clusters from david et al .
( if the actual temperature profiles would not alter their results ) . as david et al .
point out , differences in the distributions of gas and dark matter indicate that in addition to gravity , reheating and hydrodynamics must play a role in the formation of structure on cluster scales .
loewenstein & mushotzky ( 1996a ) similarly interpret the observed difference ( by a factor of @xmath114 ) of the individual baryon fractions in the central parts of a1060 and awm7 .
cluster simulations including shock heating of gas ( e.g. , pearce , thomas , & couchman 1994 ; navarro et al.1995 ) plus radiative cooling ( frenk et al .
1996 ) but not including any additional sources of gas heating , indeed produce slightly antibiased gas profiles ( @xmath115 , evrard 1997 ) due to the energy transfer from the dark matter and conversion of gas during galaxy formation .
however , they do not produce the antibias of the observed degree , suggesting that some physics is still missing from the simulations .
the most frequently mentioned candidate is energy input from supernovae ( see sarazin 1988 for a review ) . along a different line , e.g.
, david , forman , & jones ( 1991 ) and loewenstein & mushotzky ( 1996b ) showed that the supernovae needed to reproduce the elemental abundances in clusters would also produce the amount of heat comparable to the gas thermal energy .
still , the published cluster simulations that attempt to model supernovae - driven galactic winds ( e.g. , metzler & evrard 1994 ) seem to fail to reproduce the observed difference between the gas and dark matter distributions . at present , our constraints on the mass profile can be directly compared only to another result , that on a2163 ( m96a ) , since other work was limited by the unavailability of the temperature data at large radii .
the constraints on a2163 were rather similar to those obtained here ( total density profiles of the form ( 1 ) with @xmath116 were excluded , corresponding to a slightly stronger constraint on the dark matter slope ) , although they were less accurate due to a2163 s greater distance and less conclusive due to the merger in that system . using an independent method based on the analysis of ellipticity of cluster x - ray images , canizares & buote ( 1997 ) infer that the density of dark matter in clusters falls off as @xmath117 in the outer parts .
this is very similar to what we find for a2256 .
the shape of the dark matter halo may be an interesting cosmological probe ( e.g. , hoffman & shaham 1985 ; crone et al .
1994 ; see , however , navarro et al .
1997 ) . comparing a sample of just one cluster with the statistical conclusions of the simulations is uncertain and has to wait until more clusters are studied .
it is interesting to note in the meantime that the shallow density profiles predicted by kofman et al .
( 1996 ) for the chdm universe are inconsistent with our a2256 results .
gravitational lensing provides an independent measure of the cluster mass distribution ( see review in bartelmann & narayan 1995 ) .
often , masses inferred by this method are up to a factor of a few greater than those from x - rays in the central cluster regions ( e.g. , loeb & mao 1994 ; wu 1994 ; miralda - escud & babul 1995 ; tyson & fischer 1995 ) . in some ( the most discrepant ) cases , the lensing analysis is likely to overestimate the mass as a result of substructure ( bartelmann 1995 ) or projection ( daines et al.1997 ) .
bartelmann & steinmetz ( 1996 ) showed that clusters selected for the presence of strong arcs are more likely to have substructure .
however , with better temperature data , the x - ray mass estimates may also be revised upwards in better agreement with lensing measurements , as we find for a2256 .
our inferred density profiles for both functional forms are more centrally peaked than the gas profile , in agreement with lensing ( bartelmann & narayan 1995 ) and the a2163 finding ( m96a ) .
note that profiles with high central densities would erroneously be _ excluded _ in our analysis if one did not allow for the multicomponent spectrum in the central region ( required by ) and instead used a formally well - constrained single temperature that artificially excludes the possibility of a hotter gas .
this is similar to the result of allen et al .
( 1995 ) who find that allowing for a multi - temperature gas in the cooling flow region of pks0745 increases the x - ray - measured mass within the lensing radius by a large factor , such that the strong lensing and x - ray masses agree .
there are no lensing data on a2256 to compare because it is too nearby .
ikebe et al .
( 1996 , 1997 ) similarly note that central dark matter concentrations in the cd clusters fornax and hydra - a are required if the gas in their central regions is treated as multi - phase as required by the spectra .
the fact that the shallow dark matter distributions are excluded at large radii may also have implications for the weak lensing mass estimates .
these estimates at present are of a relative nature and require assumed model density profiles in order to obtain the absolute value of the mass in the central region . as squires et al .
( 1996 ) note , the central mass inferred from their analysis of a2218 varies by a factor of 1.3 depending on the assumed density profile , with the steeper profiles resulting in lower values .
our results on a2256 prefer steep density profiles , hence favor the lower bound of the weak lensing mass estimates .
regarding the dark matter distribution near the center , we note that in a cluster whose gas has a constant density core and is supported only by its own pressure , a central dark matter cusp may be ruled out a priori because it would require a convectively unstable central temperature peak .
a2256 has a central density excess over the @xmath15-model , associated with the brightest galaxy , which we do not model adequately ( for a given mass profile , it would reduce the hydrostatic central temperature ) , which is why this additional constraint was not applied here . for other clusters
, this constraint can be used to devise an interesting test of the significance of nonthermal gas support ( which was proposed as one explanation of the discrepant lensing and x - ray masses ; loeb & mao 1994 ) .
if a relaxed cluster is found whose gas indeed follows a @xmath15-model all the way to the center but whose lensing data nevertheless require a central density cusp , then the gas in the center must have nonthermal support .
we constrained the mass distribution in the primary cluster of a2256 , using actual gas temperature measurements at large radii derived from the x - ray data of unsurpassed quality for any cluster to date .
our analysis has shown that : \1 .
dark matter density profiles steeper than @xmath118 in the outer part are required by the data .
profiles with central cusps such as those predicted by simulations and implied by lensing results are consistent with the data .
the mass is distributed rather differently from what may be inferred under the assumption of isothermal gas : the total mass internal to the cluster x - ray core is higher , while the mass within the virial radius is lower , both by factors of about 1.5 , than the isothermal estimates ; \2 .
if our results for a2256 are typical for other clusters , three particularly interesting implications would be that the x - ray and lensing masses converge ; the `` baryon catastrophe '' is even more pronounced ; and gravity and merger shocks are probably not the only significant source of heating of the cluster gas .
we thank and teams for their work on obtaining the data we used here .
we are grateful to w. forman , r. mushotzky and the referees , one of whom was m. henriksen , for valuable comments on the manuscript .
this work was supported by nasa grant nag5 - 2611 and by cfa postdoctoral fellowship .
the hydrostatic equilibrium equation for a spherically symmetric cluster ( e.g. , sarazin 1988 ) , @xmath119 can be rewritten as a differential equation for the gas temperature , assuming the gas density profile of the form ( 3 ) : @xmath120 here we denote @xmath121 ; @xmath122 where @xmath70 is the gas temperature in kev at @xmath71 , so that @xmath123 ; @xmath124 and @xmath125 the two integrals in the equation above correspond to the gas and the dark matter , respectively ; @xmath126 ( where @xmath69 is the dark matter density at the x - ray core radius @xmath16 ) denotes the dimensionless dark matter density whose form is given by either ( 1 ) or ( 2 ) . the integrals in ( [ int ] ) can be worked out analytically for the density profiles we use , but it is faster to calculate them numerically directly than evaluate numerically the resulting special functions .
equation ( [ diff ] ) can be solved either numerically ( as e.g. , in hughes 1989 ) or analytically , as we do .
the analytic solution is given by @xmath127 .
\label{analyt}\ ] ] for a set of parameters @xmath68 , @xmath66 , @xmath67 , @xmath69 and @xmath70 , we calculate the temperature profile using eq .
( [ analyt ] ) .
the parameters @xmath69 and @xmath70 are adjusted so that the profile approximates the data as described in 5 . in practice , it is advantageous to fit the quantities @xmath128 and @xmath70 rather than the above two , since in the limit of zero gas mass the shape of the profile depends only on @xmath128 .
the scheme described here differs from the one used by hughes ( 1989 ) , hbn and m96a in that we parameterize a dark matter profile rather than a total mass profile .
this automatically excludes the unphysical solutions in which the dark matter density ( @xmath129 ) is negative , which required the introduction of an additional @xmath105-dependent constraint . | we combine high quality and x - ray data to constrain the radial dark matter distribution in the primary cluster of abell 2256 , free from the assumption of gas isothermality .
both instruments indicate that the temperature declines with radius .
the region including the central galaxy has a multicomponent spectrum , which results in a wide range of allowed central gas temperatures .
we find that the secondary subcluster has a temperature and luminosity typical of a rich cluster ; however , the temperature map shows no signs of an advanced merger in this double system .
it is therefore assumed that the primary cluster is in hydrostatic equilibrium .
the data then require dark matter density profiles steeper than @xmath0 in the cluster outer part .
acceptable models have a total mass within @xmath1 mpc ( approximately the virial radius ) of @xmath2 at the 90% confidence .
this is about 1.6 times smaller than the mass derived assuming isothermality .
the gas fraction is correspondingly higher and is @xmath3 .
a lower limit on the fraction of gas in the total local density at the same radius is @xmath4 , which is twice the isothermal value .
near the center , dark matter profiles with and without central cusps are consistent with the data .
our inferred total mass inside the x - ray core ( @xmath5 mpc ) is @xmath6 , which exceeds the isothermal value by a factor of 1.4 . although the confidence intervals above may be underestimates since they do not include uncertainties arising from asymmetry and departures from hydrostatic equilibrium , the behavior of the mass distribution , if applicable to other clusters , can bring into better agreement x - ray and lensing mass estimates , but aggravate the `` baryon catastrophe '' .
the observed considerable increase in the gas content with radius , not anticipated by simulations , may imply that a significant fraction of thermal gas energy comes from sources other than gravity and merger shocks , such as supernovae - driven galactic winds , for example . |
high - temperature superconductivity remains to be an exciting and rich field .
one of the interesting proposals is anderson s resonating valence bond ( rvb ) state.@xcite in the rvb theory , the parent compound is an insulator at half electron filling , or one electron per cu - site , and chemical doping is essential to introduce charge carriers to lead to superconductivity .
the mathematics of the rvb theories therefore is in a hilbert space which completely projects out the on - site double - occupied electron states . at the half
filled , there is exactly one electron per lattice site , and the charge degree of freedom is totally frozen , resulting in a mott - insulator .
another interesting class of materials in the context of strongly correlated systems is the layered organic conductors,@xcite which may undergo a phase transition from an insulator to a superconducting ( sc ) state by applying pressure.@xcite since these materials are effectively at half filling,@xcite the phase transition is due to the competition between the coulomb interaction and kinetic band width , the latter of which is tuned by pressure instead of chemical doping . there have been several related theoretical works on layered organic superconductors in recent years.@xcite the mathematics of the sc state may be described by a partially gutzwiller projected bcs state,@xcite instead of the complete projection as in the rvb theory .
we shall refer to this partially gutzwiller projected bcs state as a gossamer superconductor , a phrase first introduced by laughlin@xcite originally in the context of high temperature superconductors .
gossamer superconductivity refers to those sc states with a dilute superfluid - density .
the partial gutzwiller projection allows charge fluctuations even at half filling .
one of us@xcite proposed that in this case an effective model is the hubbard - heisenberg model which includes the standard kinetic energy , the on - site coulomb repulsion , as well as the anti - ferromagnetic spin - exchange .
the idea was applied to the study of @xmath5-(bedt - ttf)@xmath6 by gan _
et al_,@xcite , where the gutzwiller approximation was used to replace the partial gutzwiller projection by a set of renormalized factors and the resulted renormalized hamiltonian was then studied by a mean field theory .
the finding is a phase - diagram distinguishing three phases : normal metal , superconductor and anti - ferromagnet . in this paper
, we shall study the phase diagram of an effective hubbard - heisenberg model by using variational monte carlo ( vmc ) method .
the order - parameters for the d - wave superconductivity and for the anti - ferromagnetism are calculated directly .
we obtain a phase diagram consistent with the experiments , providing further support to the scenario of the gossamer superconductivity to describe the layered organic conductors .
interestingly , our numerical calculation of the sc order parameter suggests a relatively high superfluid density near the phase boundary to the afm insulator .
the results from our vmc calculations also provide support to the earlier mean field results based on the gutzwiller approximation,@xcite although we find a less sensitive role of @xmath1 .
we study a hubbard - heisenberg model in a 2-dimensional lattice illustrated in fig .
[ lattice ] .
the hamiltonian is given by @xmath7 here @xmath8 is the electron annihilation operator of an electron with spin @xmath9 on site @xmath10 , @xmath11 is the spin-1/2 operator at site @xmath10 , and @xmath12 .
the sum @xmath13 is over the nearest neighbors ( n.n . ) pairs on the square lattice , and the sum over @xmath14 is over both the n.n .
pairs and the diagonal bonds ( dashed lines in fig .
1 ) . we set the n. n. hopping @xmath15 as the energy unit , and fix the spin exchange @xmath16 as @xmath17 , and treat the diagonal hopping integral @xmath18 and the on - site repulsion @xmath2 as tuning parameters . in our numerical calculations , we consider a @xmath19 by @xmath19 lattice and use a periodic boundary condition along the x - direction and antiperiodic boundary condition along the y - direction
. our trial wave - function reads @xmath20 where @xmath21 is the bcs - wave function projected to the subspace with the fixed number of particles @xmath22 as defined in eq .
( 3 ) below . in eq . ( 2 ) , we introduce two variational parameters @xmath23 and @xmath24 , to control the partial gutzwiller projection and the afm correlation , respectively . the bcs - wave function in the fixed particle formalism
has the following form in real space : @xmath25 where @xmath26 is the spatial position of an electron with spin @xmath9 at the lattice site @xmath27 , and the sum is over all the pairs of a spin - up electron at the site @xmath27 and a spin - down electron at site @xmath28 . here @xmath29 is the amplitude of the wave function , which is the fourier transform of @xmath30 , with @xmath31 and @xmath32 given in the usual bcs wavefunction : @xmath33 we have@xcite : @xmath34 following the previous literature on the pairing symmetry for the model @xcite , we focus here on the @xmath35-wave pairing state , where @xmath36 and @xmath37 have the following forms , @xmath38 @xmath39 where @xmath40 , @xmath41 and @xmath42 are variational parameters in the theory .
note that @xmath43 in general due to the spin coupling term in the hamiltonian .
the advantage of the above trial wave function is that the sc and afm order can be treated on an equal footing .
it turns out that a small value of @xmath24 improves the energy of the sc state , while a sufficiently large value of @xmath24 leads to afm long range ordering .
we measure the staggered magnetization to quantitatively study the the afm phase , @xmath44 where @xmath45 is the number of the lattice sites , and the sum is over all the @xmath46 pairs between sites @xmath10 and @xmath47 on the lattice .
to measure the sc long range order , we introduce a pair correlation function : @xmath48 where @xmath49 is a spin singlet bond between the two sites @xmath10 and @xmath50 , and @xmath51 , @xmath52 for @xmath53 , and @xmath54 for @xmath55 . @xmath56 describes a d - wave singlet bond around the site @xmath10 .
the off - diagonal long range order parameter for the d - wave pairing can be measured by the quantity at @xmath57 , @xmath58 where the sum is over all the lattice sites . in our calculations on the finite size systems , we choose @xmath59 , the largest displacement on the lattice of @xmath19 by @xmath19 with @xmath19 upto 10 . to simplify the variational procedure , we will fix @xmath41 in the calculations with the reasons given below .
it has been argued @xcite that @xmath40 and @xmath41 are not independent in the variational calculations for the @xmath60 model .
we found that the results are essentially insensitive to @xmath41 for the present model .
in contrast , @xmath42 is an important variational parameter here .
the ground state energies and the ground state phase is sensitive to @xmath42 over a wide parameter - range . by fixing @xmath41 , we have then four variational parameters ( @xmath61 ) in our calculations to determine the phase diagram in the parameter space of @xmath2 and @xmath1 .
+ there are two sources of error bars in our numerical calculations within the variational approach .
one is from the statistical errors , and the other is due to the discreteness of the variational parameters in our calculations . in our simulation , we start with several different initial configurations and then average our numerical measurements over those simulations . the error - bars obtained in these averages are found to be one order of magnitude smaller than the error - bars described below .
we consider the possible values for the variational parameters and divide them into small slices
. then we perform vmc for all combinations of this `` mesh '' . after obtaining an optimal set of variational parameters in this mesh for a particular set of @xmath3 and @xmath4
, we develop a local mesh for nearby values of the tuning parameters . from the spacing of our mesh
, we obtain the error - bars for the variational parameters .
the results and the error - bars we present in this paper are essentially due to the finite elements we choose in the variational parameters .
of hamiltonian ( 1 ) as a function of @xmath3 for @xmath17 and various values of @xmath4 .
the inset is an enlarged figure for the energy . ] in this section , we present our results on the variational ground state , the corresponding variational parameters , the sc and afm long range orders of the hamiltonian ( 1 ) in the parameter space of @xmath3 and @xmath4 .
since the phase of the ground state is much more sensitive to the on - site repulsion @xmath2 than to the diagonal hopping @xmath1 , we focus our study on three values of @xmath1 , with @xmath62 . in fig .
[ global ] we plot the obtained ground state energies as functions of @xmath2 for different values of @xmath1 .
the corresponding optimized variational parameters and long range sc and afm order parameters as functions of @xmath2 for three sets of values of @xmath1 are plotted in fig .
[ bigpicture ] .
the simulations are carried out on lattice sizes of l=6 , 8 , and 10 , as indicated in the figure .
before we discuss the results , we note that the spin coupling term in eq .
( 1 ) is to account the virtual hopping process in the hubbard model , which is derived at the large @xmath2 limit .
the present study may be of relevance to the hubbard model only at large @xmath2 , but not at small @xmath2 .
our main interest will be at large or intermediate values of @xmath2 , and the interpretation of our results at small @xmath2 to the hubbard model should be cautious . before we discuss general features , we briefly discuss the obtained variational parameter @xmath42 , which is to optimize the kinetic energy due to the presence of the diagonal hopping integral @xmath1 .
@xmath42 increases as @xmath1 increases , but @xmath42 is significantly smaller than @xmath1 as we can see from the first row in fig .
[ bigpicture ] . at large @xmath2 , @xmath42 becomes zero or very tiny .
this may be understood as a result of the afm ground state with commensurate wave vector @xmath63 , since a finite @xmath42 does not match the afm state and is not preferred .
as @xmath2 increases from zero , the projection parameter @xmath23 increases from around @xmath64 , indicating a graduate increase in gutzwiller projection , while the weighting factor parameter @xmath24 changes little at small @xmath2 , but changes rapidly around @xmath65 .
the mean field pairing amplitude parameter @xmath40 changes slowly at small @xmath2 , but increases rapidly starting from around @xmath66 , then reaches a maximum at around @xmath67 and drops at larger @xmath2 .
the ground state properties are best seen in the measurement of the sc order parameter @xmath68 and afm order parameter @xmath69 .
qualitatively there are three regions as @xmath2 increases . at small @xmath2 , both @xmath68 and @xmath69
are very tiny or essentially zero , indicating a metallic state . at intermediate @xmath2 ,
@xmath68 increases monotonically with @xmath2 , while @xmath69 remains tiny , indicating a sc state .
as @xmath2 further increases , @xmath68 drops sharply , while @xmath69 increases rapidly .
we may identify this phase as the afm phase without the sc order .
the tiny but non - zero values of @xmath68 and @xmath69 in the non - ordered states may be explained as finite size effect , although a systematic scaling analyses is difficult due to the small sizes we have studied .
the above features are qualitatively similar for @xmath70 .
this is somewhat different from the early analytic calculations by using gutzwiller approximations on the projected wavefunctions , where @xmath1 is found to suppress the afm phase .
we note that while the onset for the sc - phase is similar for different @xmath1 , the magnitude of the sc - order parameter is much bigger for the @xmath71 and @xmath72 .
as we can see from the figure , the largest sc order parameter @xmath68 is found near the boundary to the afm - phase . at @xmath73
, we expect the model ( 1 ) to have instability towards a commensurate afm state for any finite @xmath2 . in fig .
[ gossphase ] , we plot the phase diagram of model ( 1 ) obtained within our variational wavefunctions .
while the phase - boundary between sc and afm can be found easily by considering one point clearly belonging to the afm and one point clearly belonging to the sc phase , between sc and metallic phase we have to use an arbitrary value to define the phase - boundary , as the onset of the sc - order parameter @xmath68 is not so sharp .
we choose @xmath74 as our criteria classifying the phase to be sc .
the error - bars in this diagram reflect within which area we have uncertainty that a point would be in either of the two phases considered . comparing the phase diagram obtained in the vmc method with the previous result by using renormalized mean field theory,@xcite they qualitatively agree with each other in the sense both give the three phases , and overall features are similar . however , there are two differences . first , while both of the methods give the transition point between the sc- and afm - phases to increase when @xmath4 increases , in the vmc calculation , the effect is not as big as in the earlier gutzwiller - approximation based calculation . in our calculation
if we consider a fixed and non - zero @xmath42 instead of a variational one , we would in fact get a slope close to the one reported by gan _
et al_. the second difference is that our vmc suggests the onset for superconductivity to be at @xmath75 for all cases , while gan _
et al _ find this phase boundary changing considerable when tuning @xmath4 .
we believe that these differences can be attributed to the different method and wave - function used . for comparison with the experiments ,
we plot a schematic phase diagram for the layered organic conductors at the right panel of the figure . in summary ,
we have presented the results of vmc calculations for a recently suggested model for gossamer superconductivity .
our trial wave function has the ingredient to describe metallic , afm and sc states .
this was archived by means of using jastrow - factors for partial gutzwiller - projection and afm - weighting .
we showed that the vmc result is consistent with experiments , and supports the previously suggested analytical variational calculations qualitatively , as we were able to identify the three expected phases , with the help of measurements of the order - parameters for afm and sc .
the exact transition line between sc- and metallic - phase , and between sc- and afm - phase differs from the one found previously .
the vmc calculations have been carried out on the hpcpower - cluster , the winhpc - cluster and the windows - condor all powered by hku s computer centre .
we wish to thank masao ogata for advise on vmc .
further we wish to thank kwan wing keung for help in parallelization of the code , and customization of the code for the windows - condor - system .
the work was partly supported by hong kong s rgc grant .
the work in nanjing was supported by nsfc 10325416 , the ministry of science and technology of china ( under the grant no .
2006cb921802 and 2006cb601002 ) and the 111 project ( under the grant no . | we use a partially gutzwiller projected bcs d - wave wavefunction with an antiferromagentic weighting factor to study the ground state phase diagram of a half filled hubbard - heisenberg model in a square lattice with nearest neighbor hopping @xmath0 and a diagonal hopping @xmath1 .
the calculations are carried out by using variational monte carlo method which treats the gutzwiller projection explicitly . at large on -
site coulomb interaction @xmath2 , the ground state is antiferromagnetic .
as @xmath2 decreases , the ground state becomes superconducting and eventually metallic .
the phase diagram is obtained by extensive calculations . as compared to the strong effect of @xmath3 , the phase boundaries turn out to be less sensitive to @xmath4 .
the result is consistent with the phase diagram in layered organic conductors , and is compared to the earlier mean field result based on the gutzwiller approximation . |
blazars constitute the most extreme subclass of active galactic nuclei ( agn ) , characterized by the emission of strong non - thermal radiation across the entire electromagnetic spectrum and in particular intense @xmath1-ray emission above 100 mev .
the typical observational properties of blazars include irregular , rapid , high - amplitude variability , radio - core dominance , apparent super - luminal motion , a flat radio spectrum , and high and variable polarization at radio and optical frequencies .
these features are interpreted as resulting from the emission of high - energy particles accelerated within a relativistic jet closely aligned with our line of sight and launched in the vicinity of the supermassive black hole harboured by the active galaxy @xcite . since the advent of the energetic gamma ray experiment telescope ( egret ) on the compton gamma - ray observatory ( cgro )
, blazars were known to dominate the extragalactic high - energy sky .
however , egret did not pinpoint the location of many sources with sufficient precision to enable astronomers to associate them with known objects , leaving the legacy of a large fraction of unidentified sources in @xmath1 rays .
the point - spread function and sensitivity of the large area telescope ( lat ) on board _ fermi _ provides an unprecedented angular resolution at high energies for localizing a large number of newly found @xmath1-ray emitting sources .
correlated variability observed at different frequencies can give important information for the identification of a @xmath1-ray source with its low - energy counterpart . on 2011 december 14 a @xmath1-ray flare from a source positionally consistent with pks 2123@xmath0463 was detected by _
fermi_-lat @xcite , triggering grond and _ swift _ follow - up observations @xcite that confirmed contemporaneous activity in the optical / uv as well as marginally in x - rays .
pks 2123@xmath0463 is a bright radio quasar with a luminosity at 1.4 ghz l@xmath8= ( [email protected])@xmath1010@xmath11 w / hz ( assuming the redshift estimated in this paper , @xmath12 = [email protected] = [email protected] is d@xmath13 = [email protected] gpc and 1 arcsec corresponds to a projected distance of [email protected] kpc . ] ; see sect .
[ redshift ] ) . on the basis of its spectral index @xmath140.4 ( s(@xmath15 ) @xmath16 ) between 408 mhz and 4.8 ghz ,
it was included in the crates catalog of flat spectrum objects @xcite .
atca observations performed almost simultaneously at 4.8 , 8.6 and 20 ghz during the australia telescope 20 ghz ( at20 g ) survey indicated a flattening of the radio spectrum at high frequencies to a spectral index of @xmath17 .
polarized emission has been detected only at 4.8 ghz where the polarization is about 4% of the total intensity flux density @xcite .
observation of pks 2123@xmath0463 in march 2004 has shown the presence of a jet - like extended structure in x - rays @xcite .
the redshift of pks 2123@xmath0463 was reported to be @xmath3 @xcite based on an objective - prism spectrum , subsequently questioned by @xcite because two possible redshifts ( 0.48 and 1.67 ) were given from observation of two lines in the spectrum , and the motivation for the exclusion of the smaller redshift was not provided .
this object was a member of the pre-_fermi _ launch roma - bzcat @xcite catalog listing candidate @xmath1-ray blazars detectable by _ fermi_-lat but not in the cgrabs catalog @xcite .
the flat spectrum radio quasar ( fsrq ) pks 2123@xmath0463 ( r.a.:21h26m30.7042s , dec.:@xmath046d05m47.892s , j2000 ; * ? ? ?
* ) was associated in the first _
fermi_-lat source catalog ( 1fgl , 2008 august 4 2009 july 4 ; * ? ? ?
* ) with the @xmath1-ray source 1fgl j2126.1@xmath04603 , while no association with a @xmath1-ray source was present in the second _ fermi_-lat source catalog ( 2fgl , 2008 august 4 2010 august 4 ; * ? ? ?
* ) , although a @xmath1-ray source , 2fgl j2125.0@xmath04632 with a radius of the 95% source location confidence region of 0.17@xmath18 , at 0.52@xmath18 from the radio position of pks 2123@xmath0463 was reported .
taking into consideration the high variability of blazars , the flux of pks 2123@xmath0463 could have decreased in the second year of _ fermi_-lat operation .
in addition , in the construction of the 2fgl catalog , pks 2123@xmath0463 was split into more than one candidate source seed ( see section 4.2 of * ? ? ? * for details ) .
these two factors may have led to the lack of its association with a @xmath1-ray source .
in this paper we present the localization over 41 months of _ fermi_-lat data of a @xmath1-ray source associated with the fsrq pks 2123@xmath0463 .
the correlated variability observed in optical , uv , x - ray and @xmath1 rays confirms the identification .
in addition a new estimation of the redshift of the source by means of the fit of simultaneous grond and _ swift_/uvot data collected soon after the @xmath1-ray flare is presented
. the paper is organized as follows : in section 2 we report the lat data analyses and results . in sections 3 and 4
we report the results of the _ swift _ and grond data analysis , respectively .
radio data collected by the atca , ceduna and kat-7 telescopes are presented in section 5 .
section 6 presents an estimation of the redshift of the source . in section 7
we discuss the modeling of the overall sed and draw our conclusions . throughout the paper a @xmath19cdm cosmology with h@xmath20 = 71 km s@xmath5 mpc@xmath5 , @xmath21 = 0.73 , and @xmath22 = 0.27 is adopted .
the _ fermi_-lat is a @xmath1-ray telescope operating from @xmath23mev to @xmath24gev .
it has a large peak effective area ( @xmath25 @xmath26@xmath27 for @xmath28gev photons ) , an energy resolution of typically @xmath29 , and a field of view of about @xmath30 sr with an angular resolution ( @xmath31 containment angle ) better than 1@xmath18 for energies above @xmath28gev .
further details about the _
fermi_-lat are given by @xcite .
100 mev ) light curve of pks 2123@xmath0463 obtained from 2008 august 4 to 2012 january 4 ( mjd 5468255930 ) .
the last bin covers 1 month of observation .
arrows refer to 2-@xmath32 upper limits on the source flux .
upper limits are computed when ts @xmath33 10.,width=264 ] the _
fermi_-lat data reported in this paper were collected over the first 41 months of _ fermi _ operation , from 2008 august 4 ( mjd 54682 ) to 2012 january 4 ( mjd 55930 ) . during this time the lat instrument operated almost entirely in survey mode .
the spectral analysis was performed with the ` sciencetools ` software package version v9r23p1 .
fermi_-lat data were extracted from a circular region of interest ( roi ) with a @xmath34 radius centred at the radio location of pks 2123@xmath0463 .
only events belonging to the ` source ' class were used .
the time intervals when the rocking angle of the lat was greater than 52@xmath18 were rejected .
in addition , a cut on the zenith angle ( @xmath35 ) was also applied to reduce contamination from the earth limb @xmath1 rays , which are produced by cosmic rays interacting with the upper atmosphere .
the spectral analyses ( from which we derived spectral fits and photon fluxes ) were performed with the post - launch instrument response functions ( irfs ) ` p7source_v6 ` using an unbinned maximum likelihood method implemented in the science tool ` gtlike ` .
100 mev ) light curve of pks 2123@xmath0463 obtained from 2011 december 4 to 2012 january 4 ( mjd 5589955930 ) with 2-day or 4-day time bins .
arrows refer to 2-@xmath32 upper limits on the source flux .
upper limits are computed when ts @xmath33 10.,width=264 ] the background model used to extract the @xmath1-ray signal includes a galactic diffuse emission component and an isotropic component .
the model that we adopted for the galactic component , the same as used for the 2fgl catalog , is given by the file gal_2yearp7v6_v0.fits , and the isotropic component , which is the sum of the extragalactic diffuse emission and the residual charged particle background , is parametrized by the file iso_p7v6source.txt .
the normalizations of both components in the background model were allowed to vary freely during the spectral point fitting .
we examine the significance of the @xmath1-ray signal from the sources using the test statistic ( ts ) based on the likelihood ratio test .
the test statistic , defined as ts = 2@xmath36log(likelihood ) between models with and without the source , is a measure of the probability of having a @xmath1-ray source at the localization specified , which compares models whose parameters have been adjusted to maximize the likelihood of the data given the model @xcite . the source model used in ` gtlike `
includes all the point sources from the 2fgl that fall within @xmath37 of pks 2123@xmath0463 .
the spectra of those sources were parametrized by power - law functions , except for 2fgl j2056.2@xmath04715 for which we used a log - parabola , and 2fgl j2124.6@xmath03357 and 2fgl j2241.7@xmath05236 for which we used an exponentially cut - off power - law in their spectral modeling as in the 2fgl catalog .
we removed from the model the sources having ts @xmath33 25 and/or fluxes ( 0.1100 gev ) below 1.0@xmath1010@xmath38 photons cm@xmath5 s@xmath5 over 41 months and repeated the fit .
we tested whether two distinct @xmath1-ray sources ( one at the radio position of pks 2123@xmath0463 and one at the @xmath1-ray position of 2fgl j2125.0@xmath04632 ) are detected simultaneously by _ fermi _ with ts @xmath40 25 over 41 months of observations . in this scenario ,
the fit yields a ts of 553 for pks 2123@xmath0463 and a ts of 14 for 2fgl j2125.0@xmath04632 .
we thus conclude that pks 2123@xmath0463 and 2fgl j2125.0@xmath04632 are the same source and we maintain only pks 2123@xmath0463 in the model .
thus a final fitting procedure has been performed with all sources within 10@xmath18 of pks 2123@xmath0463 included with the normalization factors and the photon indices left as free parameters .
for the sources located between 10@xmath18 and 15@xmath18 we kept the normalization and the photon index fixed to the values obtained in the previous fitting procedure .
the roi model includes also sources falling between 15@xmath18 and 20@xmath18 from the target source , which can contribute to the total counts observed in the roi due to the energy - dependent size of the point spread function of the instrument . for these additional sources , normalizations and indices
were fixed to the values of the 2fgl catalog during all steps of the fitting procedure .
the @xmath1-ray point source localization using the ` gtfindsrc ` tool over the photons extracted during the period 2008 august 4 2012 january 4 results in r.a .
= 321.609@xmath18 , dec .
= @xmath046.076@xmath18 ( j2000 ) , with a 95@xmath41 error circle radius of 0.047@xmath18 , at an angular separation of 0.024@xmath18 from the radio position of pks 2123@xmath0463 ( r.a .
= 321.628@xmath18 , dec .
= @xmath046.097@xmath18 , j2000 ) .
this implies a strict spatial association with pks 2123@xmath0463 .
the fit with a power - law ( pl ) model , dn / de @xmath42 ( e / e@xmath43)@xmath44 , to the data integrated over 41 months of _ fermi _ operation in the 0.1100 gev energy range results in a ts = 557 , with an integrated average flux of ( 4.76 @xmath9 0.35 ) @xmath1010@xmath38 photons @xmath45 s@xmath5 , and a photon index @xmath46 = 2.45 @xmath9 0.05 ( with e@xmath20 fixed to 670 mev ) . in order to test for curvature in the @xmath1-ray spectrum of pks 2123@xmath0463 we used as an alternative spectral model the log - parabola ( lp ) , dn / de @xmath42 ( e / e@xmath43)@xmath47 @xcite , where the parameter @xmath48 is the spectral slope at the energy e@xmath20 and the parameter @xmath49 measures the curvature around the peak .
we fixed the reference energy e@xmath20 to 300 mev .
the fit with a log - parabola results in a ts = 554 , with spectral parameters @xmath48 = 2.35 @xmath9 0.09 , and @xmath49 = 0.06 @xmath9 0.04 . applying a likelihood ratio test to check the pl model ( null hypothesis ) against the lp model ( alternative hypothesis ) , the power - law spectral model is favoured , indicating that no significant curvature was observed in the average @xmath1-ray spectrum .
figure [ fig1 ] shows the @xmath1-ray light curve of the first 41 months of _ fermi _ observations of pks 2123@xmath0463 built using 2-month time bins , with the exception of the final ( 1-month ) data point . for each time
bin the photon index was frozen to the value resulting from the likelihood analysis over the entire period .
if ts @xmath33 10 2-@xmath32 upper limits were calculated instead .
the systematic uncertainty in the flux is energy dependent : it amounts to @xmath50 at 100 mev , decreasing to @xmath51 at 560 mev , and increasing to @xmath50 above 10 gev @xcite . as shown in fig .
[ fig1 ] pks 2123@xmath0463 was in a low @xmath1-ray state ( 0.1100 gev flux @xmath33 5@xmath1010@xmath38
ph @xmath45 s@xmath5 ) for the first 2.5 years . in particular the source
was not detected with 2-month time bin during the period 2009 august2010 may .
a first increase of the @xmath1-ray flux was observed in 2011 february - march , and subsequently a flaring activity in 2011 december .
we show a light curve focused on the period of high activity ( 2011 december 42012 january 4 ; mjd 5589955930 ) with 2-day or 4-day time bins ( fig .
[ fig2 ] ) .
the peak of the emission was observed between december 13 15:43 ut and december 14 15:43 ut , with a flux of ( 128 @xmath9 23 ) @xmath1010@xmath38 photons @xmath45 s@xmath5 in the 0.1100 gev energy range , a factor of @xmath25 25 higher with respect to the average @xmath1-ray flux observed during 2008 - 2011 .
the corresponding observed isotropic @xmath1-ray luminosity peak in the 0.1100 gev energy range is [email protected] @xmath10 10@xmath52 erg s@xmath5 ( assuming @xmath12 = [email protected] , see sect .
[ redshift ] ) , comparable to the values reached by the most powerful fsrqs in flaring activity ( e.g. * ? ? ?
* ; * ? ? ?
leaving the photon index free to vary during the period of high activity and @xmath53 fixed to 670 mev the fit results in a photon index @xmath46 = 2.26 @xmath9 0.06 , showing a moderate harder - when - brighter behaviour already observed in other fsrqs @xcite .
replacing in the same period the pl with a lp , fixing the reference energy e@xmath20 to 300 mev , we obtain spectral parameters @xmath48 = 2.09 @xmath9 0.13 , and @xmath49 = 0.10 @xmath9 0.04 .
we used a likelihood ratio test to check the pl model ( null hypothesis ) against the lp model ( alternative hypothesis ) .
these values may be compared , following @xcite , by evaluating the curvature test statistic ts@xmath54 = 2(log(like)@xmath55-log(like)@xmath56)=3 ( @xmath251.7-@xmath32 ) ; thus no significant curvature was observed in the @xmath1-ray spectrum also during the high activity period .
finally , using the ` gtsrcprob ` tool , which calculates the probabilities an event belongs to each of the sources in the model , we estimated that the highest - energy photon emitted by pks 2123@xmath0463 was observed with a probability of 87% at distance of 0.35@xmath18 from the source with an energy of 92 gev .
for a single high - energy photon ( @xmath40 1 gev ) , sky location can only be determined within @xmath25 0.8@xmath18 ( 95% confidence radius ) .
the _ swift _ satellite @xcite performed two observations of pks 2123@xmath0463 in 2011 december triggered by the @xmath1-ray flare detected by
_ fermi_-lat .
as a comparison we also analysed two additional observations performed during 20102011 .
the observations were performed with all three on - board instruments : the x - ray telescope ( xrt ; * ? ? ?
* kev ) , the ultraviolet optical telescope ( uvot ; * ? ? ?
* nm ) and the burst alert telescope ( bat ; * ? ? ?
* kev ) .
the hard x - ray flux of this source is below the sensitivity of the bat instrument for the short exposure time of these observations .
moreover , the source was not present in the _ swift _ bat 58-month hard x - ray catalog @xcite and the 54-month palermo bat catalog @xcite .
[ cols="^,^,^,^,^ " , ]
the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden .
additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national dtudes spatiales in france .
part of the funding for grond ( both hardware as well as personnel ) was generously granted from the leibniz - prize to prof .
g. hasinger ( dfg grant ha 1850/28 - 1 ) .
we thank the swift team for making these observations possible , the duty scientists , and science planners .
dak acknowledges support by dfg grant kl 766/16 - 1 , and is grateful for travel funding support through mpe .
the australia telescope compact array is part of the australia telescope national facility which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
this research was funded in part by nasa through fermi guest investigator grant nnh09zda001n ( proposal number 31263 ) .
this research was supported by an appointment to the nasa postdoctoral program at the goddard space flight center , administered by oak ridge associated universities through a contract with nasa .
we thank silvia rain and the anonymous referee for useful comments and suggestions .
stevens , j. a. , et al .
2012 , proceedings of fermi & jansky : our evolving understanding of agn , st michaels , md , november 10 - 12 , 2011 , edited by r. ojha , d. j. thompson and c. dermer , econf c1111101 , arxiv:1205.2403 | the flat spectrum radio quasar ( fsrq ) pks 2123@xmath0463 was associated in the first _
fermi_-lat source catalog with the @xmath1-ray source 1fgl j2126.1@xmath04603 , but when considering the full first two years of _ fermi _ observations , no @xmath1-ray source at a position consistent with this fsrq was detected , and thus pks 2123@xmath0463 was not reported in the second _ fermi_-lat source catalog . on 2011 december 14 a @xmath1-ray source positionally consistent with pks 2123@xmath0463 was detected in flaring activity by _
fermi_-lat .
this activity triggered radio - to - x - ray observations by the _
swift _ , grond , atca , ceduna , and kat-7 observatories .
results of the localization of the @xmath1-ray source over 41 months of _ fermi_-lat operation are reported here in conjunction with the results of the analysis of radio , optical , uv and x - ray data collected soon after the @xmath1-ray flare .
the strict spatial association with the lower energy counterpart together with a simultaneous increase of the activity in optical , uv , x - ray and @xmath1-ray bands led to a firm identification of the @xmath1-ray source with pks 2123@xmath0463 .
a new photometric redshift has been estimated as @xmath2 using grond and _ swift_/uvot observations , in rough agreement with the disputed spectroscopic redshift of @xmath3 .
we fit the broadband spectral energy distribution with a synchrotron / external compton model .
we find that a thermal disk component is necessary to explain the optical / uv emission detected by _ swift_/uvot .
this disk has a luminosity of @xmath4 erg s@xmath5 , and a fit to the disk emission assuming a schwarzschild ( i.e. , nonrotating ) black hole gives a mass of @xmath6 m@xmath7 .
this is the first black hole mass estimate for this source .
galaxies : active galaxies : nuclei quasars : general quasars : individual ( pks 2123 - 463 ) gamma rays |
establishing a quantum theory of gravity ( also called quantum gravity ) is one of the main challenges in modern physics . to gain insights into the development of such quantum theories , it is useful to investigate experimentally accessible quantum gravity effects . in general , such effects are very small , expected to be inversely proportional to the planck energy scale @xmath3 gev .
even so , current experimental results , mainly from high - energy physics , astrophysics , and cosmology , can set meaningful bounds on parameters relevant to models on quantum gravity
. a lot of work has been done , and references can be found in a recent review paper @xcite . here , we are interested in one particular quantum gravity effect , the so - called generalized uncertainty principle ( gup ) . according to the heisenberg uncertainty principle ( hup ) , the uncertainties in the measurement of the length and the momentum satisfy the relation @xmath4\rangle|=\hbar/2 $ ] .
in other words , the uncertainty @xmath5 is bounded by @xmath6 .
therefore , on the cost of the information of the momentum @xmath7 , the length can be arbitrarily precisely measured . on the other hand
, it has long been known that various models on quantum gravity predict the existence of a minimum measurable length @xcite .
thus , this forces people to modify the hup to the gup .
the gup has various implications on a wide range of physical systems , especially in the quantum regime .
for example , effects of the gup on the thermodynamics of the quark - gluon plasma were studied in ref .
the impact of the gup on thermodynamical parameters and the stability of the schwarzschild black hole was investigated in ref .
the gup approach was also used to calculate quantities of the inflationary dynamics and the thermodynamics of the early universe @xcite .
the violation of the equivalence principle due to the gup effects was discussed in refs .
furthermore , it was suggested that experimental data of the lamb shift , the landau levels , and the tunneling current in a scanning tunneling microscope could be used to constrain the gup corrections @xcite .
a direct measurement of the gup effects in quantum optics laboratory was proposed in ref . @xcite .
more references can be found in ref .
@xcite . in recent years
, rapid technological advances in atom interferometry have been achieved . due to their high sensitivity , atom interferometers
have already been used in various precision measurements , and many important experimental results came out .
for example , the value of the newtonian gravitational constant was measured to be @xmath8 @xcite , which just differs by 1.5 combined standard deviations from the codata recommended value @xcite .
a @xmath9-level test of the weak equivalence principle ( wep ) was reported in a recent work @xcite .
atom interferometers can also be used to search deviations from newton s @xmath10 law at the micrometer scale @xcite .
another important application of atom interferometers in precision measurements is the determination of the fine structure constant , @xmath11 .
a value of @xmath12=137.035999037(91 ) was obtained @xcite , which is the second best value compared to the one deduced from the electron anomaly measurement @xcite .
more details on atom interferometers can be found in many review papers , such as ref .
@xcite . inspired by these achievements with atom interferometry
, people began to think about the possibility of studying quantum gravity effects on atomic physics .
one such study was given in ref .
@xcite , where the authors used the atom recoil measurement @xcite to constrain parameters in the energy - momentum dispersion relation modified by quantum gravity . in this paper , we investigate the possibility of using the cold atom recoil measurement to constrain parameters in three popular gup proposals .
improved bounds on these parameters are found .
the paper is organized as follows . in sect .
ii , three popular gup proposals are introduced .
then a brief description of the cold atom recoil experiment is given in sect .
iii . through a detailed calculation of the gup effects on the cold atom recoil measurement , bounds on the gup parameters are obtained in sect .
finally , we conclude the paper in sect . v.
as discussed above , various models on quantum gravity such as string theory and loop quantum gravity suggest the existence of a minimum measurable length .
consequently , this indicates that the hup should be modified to the gup at energies close to the planck energy scale @xmath13 .
but at the moment no model has the power of making robust predictions on what the gup should be .
the alternative way is making various proposals on the gup .
depending on what indications from models on quantum gravity should be incorporated , many different gup proposals have been made @xcite . here , we discuss three popular ones .
first , let us consider the so - called kempf - mangano - mann ( kmm ) proposal , which was first discussed in ref .
the authors were motivated by the observation that a variety of models of quantum gravity predicted a leading quadratic - in - the - momenta - type correction to the hup . to incorporate this observation ,
the following form is proposed : @xmath14=i \hbar \left(\delta_{ij } + \frac{\beta_0}{(m_p c)^2 } \delta_{ij } p^2 + \frac{2\beta_0}{(m_p c)^2 } p_i p_j\right ) , \label{kmm}\ ] ] where @xmath15 is a dimensionless parameter , and @xmath16 is the planck mass .
all other commutation relations vanish .
it is easy to find the following uncertainty relation @xmath17 where @xmath18 .
this inequality relation implies a minimum measurable length @xmath19 , where @xmath20 is the planck length .
theoretically , the planck length is believed to be the minimal measurable length .
thus , @xmath15 is normally assumed to be of the order of unity .
however , if one does not take the above assumption _ a priori _ , current experiments can set upper bounds on it .
for example , the standard model of high - energy physics is well tested at energy scale 100 gev , which implies that @xmath21 .
better bounds are obtained in ref .
@xcite , where the best one @xmath22 is set by the tunneling current measurement .
next , we discuss the ali - das - vagenas ( adv ) proposal , which was first put forward in ref .
the authors observed that doubly special relativity theories @xcite suggested a leading linear - in - the - momenta type correction to the hup . to incorporate both the linear and the quadratic in - the - momenta corrections , they proposed the following form : @xmath23&= & i \hbar \left(\delta_{ij } - \frac{\eta_0}{m_p c } ( \delta_{ij } p+ p_i p_j / p)\right.\\ & & \left .
+ \frac{\eta_0 ^ 2}{(m_p c)^2}(\delta_{ij } p^2 + 3 p_i p_j)\right),\end{aligned}\ ] ] where @xmath24 is a dimensionless parameter , and is normally assumed to be of the order of unity .
all other commutators vanish . again ,
if the assumption on @xmath24 is not taken _ a priori _ , one can use current experiments to set upper bounds on it .
the minimum measurable length of this proposal is easy to find to be @xmath25 .
then , an immediate bound from high - energy physics is @xmath26 .
the best bound is set by the lamb shift measurement @xcite , which is @xmath27 .
the last proposal to be discussed is maggiore s proposal @xcite , which is motivated by the relationship between the gup and the quantum deformation of the poincar@xmath28 algebra .
the author gave the proposal @xmath14=i \hbar \ , \delta_{ij}\sqrt{1+\frac{\gamma_0}{(m_p c)^2}(p^2+m^2 c^2 ) } , \label{maggiore1}\ ] ] where @xmath29 is a dimensionless parameter , and is normally assumed to be of the order of unity .
if this assumption is not taken _ a priori _ , current experiments can be used to set upper bounds on it .
the minimum measurable length of this proposal is found to be @xmath30 .
again , a direct bound from high - energy physics is @xmath31 . currently , no other bounds are available .
since @xmath32 is very small , one can make taylor series expansion on eq .
( [ maggiore1 ] ) @xmath14=i \hbar \ , \delta_{ij}\left(1+\frac{\gamma_0}{2(m_p c)^2}(p^2+m^2 c^2)\right).\ ] ] compared to the kmm proposal , the above formula is very similar to it , and can be regarded as a generalization of it .
but the difference will be important later .
type atomic interferometer combined with @xmath33 bloch oscillations.,width=400 ] a typical cold atom recoil experiments can be described with the schematic diagram shown in fig .
1 , where two hyperfine ground states of atoms are labeled by @xmath34 and @xmath35 , respectively . it combines a ramsey - bord@xmath36-type atomic interferometer with bloch oscillations so that as many as possible recoils can be coherently transferred to atoms and atomic velocities can be accurately measured with the raman transition between two hyperfine ground states .
details of the experimental setup can be found in ref .
the experimental procedure can be described as follows .
a slow atomic beam , prepared in the @xmath34 state by a two - dimensional magneto - optical trap ( 2d - mot ) , is loaded into the interferometer through a standard three - dimensional magneto - optical trap ( 3d - mot ) .
after a few seconds of cooling , a first pair of raman @xmath37/2 pulses , realized by using two counterpropagating laser beams with frequencies @xmath38 and @xmath39 , and wave vectors @xmath40 and @xmath41 , is applied to create two coherent atomic beams in the @xmath35 state , and define an initial velocity distribution centered on @xmath42 .
then , atoms remaining in the @xmath34 state are blown away by a resonant laser beam tuned to the single photon transition .
this process is called velocity selection of atoms . at resonance , according to the energy conservation law , we have @xmath43 where @xmath44 is the mass of atoms , @xmath45 is the single photon detuning of the atomic levels , and @xmath46 is the energy difference between the @xmath34 state and the @xmath47 state .
this equation can be written as @xmath48 where @xmath49 is defined to be the raman detuning from the atomic resonance .
after the velocity selection , a number of @xmath33 bloch oscillations are applied to both atomic beams .
bloch oscillations are an efficient way of transferring very high recoil velocities to atoms in a short time , while leaving the internal state of atoms unchanged . by using two counterpropagating laser beams ,
atoms are coherently accelerated through a succession of two - photon raman transitions . in each bloch oscillation , the atomic velocity increases by 2@xmath50 , where @xmath51 is the recoil velocity of the atom when absorbing a photon of momentum @xmath52 .
thus , a final velocity , @xmath53 , is achieved for atoms .
finally , a second pair of raman @xmath37/2 pulses is applied to read out the final velocity distribution centered on @xmath54 , and transfer the atoms back to the @xmath34 state . again , according to the energy conservation law , the raman detuning for the velocity measurement is found to be @xmath55 subtracting eq .
( [ sel ] ) from eq .
( [ meas ] ) , one can easily find that @xmath56 then , the ratio @xmath57 is determined by @xmath58 where @xmath59 is the planck constant .
more importantly , the fine structure constant @xmath11 can be deduced from the value of @xmath57 with the well - known relation @xmath60 where @xmath61 is the rydberg constant , @xmath62 is the speed of light , and @xmath63 is the electron mass . the best measurement of @xmath57 was given in ref .
@xcite , where the 5s@xmath64 @xmath65f=2 , m@xmath66=0@xmath67 and 5s@xmath64 @xmath65f=1 , m@xmath66=0@xmath67 states of @xmath0rb atoms were adopted as @xmath34 and @xmath35 , respectively .
the atomic beam was prepared in the @xmath68 state with an initial velocity @xmath69 m / s .
the bloch oscillations were generated by a ti : sapphire laser with wavelength @xmath70 nm .
a number of @xmath71 bloch oscillations were applied to @xmath0rb atoms in each run . after a careful analysis of error budget ,
the value of @xmath72 was measured to be @xmath73 in the next section , we will discuss how to use the above measurement to constrain parameters in previous three gup proposals .
following the procedure in ref .
@xcite , we make redefinitions in the kmm proposal ( [ kmm ] ) , @xmath74 where @xmath75 .
it is easy to check that @xmath76 and @xmath77 now satisfy the canonical commutation relations @xmath78=i \hbar \ , \delta_{ij}$ ] .
thus , it is natural to interpret @xmath79 as momentum at a low - energy scale , i.e. , @xmath80 in position space .
@xmath81 can be regarded as momentum at a high - energy scale .
consider a quantum system with the hamiltonian @xmath82 with the redefinitions ( [ redef ] ) , the above hamiltonian can be rewritten as @xmath83 where @xmath84 higher order terms are omitted .
next we apply the gup - corrected hamiltonian [ eq .
( [ gupkin ] ) ] to the cold @xmath0rb atoms .
it turns out that the gup effects modify the kinetic energy of @xmath0rb atoms into @xmath85 repeating the previous derivation , we find a gup - corrected relation : @xmath86 combining this equation with eq .
( [ alpha ] ) , one can get @xmath87 where @xmath88 is the atomic mass unit .
note that @xmath61 , @xmath11 , @xmath89 , and @xmath90 are all measured with very high accuracy . with values in table i , and the measurement ( [ hmeas ] )
, we finally find that @xmath91 this bound on @xmath15 is weaker than those set by high - energy physics and measurements of the lamb shift , and better than the one from measurements of landau levels @xcite .
the reason is that @xmath15 is always associated with the factors , @xmath92 and @xmath93 , in the calculation .
even if heavier atoms are chosen , the factor @xmath92 could not be substantially greater than @xmath94 .
since we are talking about cold atoms , the factor @xmath93 is at most of order @xmath95 .
taking these two limitations into consideration , it is hopeless for cold atom recoil experiments to give a better bound on @xmath15 than the one from high - energy physics .
but things are different for the adv proposal .
.quantities used in our calculation [ cols="<,<,^,^",options="header " , ] similar to the case of the kmm proposal , make redefinitions in the adv proposal : @xmath96 now @xmath76 and @xmath77 satisfy the canonical commutation relations @xmath78=i \hbar \ , \delta_{ij}$ ] .
then , taking the redefinitions ( [ advred ] ) to the hamiltonian eq .
( [ hamiltonian ] ) , we have @xmath97 we the apply this gup - corrected hamiltonian to the cold @xmath0rb - atom system .
the kinetic energy of @xmath0rb atoms is modified by the gup effects as @xmath98 where only the linear term in @xmath99 is kept . by a similar derivation
, one can get a gup - modified relation @xmath100 using eq .
( [ alpha ] ) , we find that @xmath101 the measurement ( [ hmeas ] ) and quantities in table i determine an upper bound @xmath102 this bound on @xmath24 is better than those set by high - energy physics and measurements of landau levels , and weaker than the one from measurements of the lamb shift @xcite .
note that @xmath24 is always associated with the factors , @xmath103 and @xmath104 , in the calculation .
compared to the case of the kmm proposal , effects from the adv proposal are larger .
better bounds can be found in the future .
but it is challenging to set an order - of - unity bound on @xmath24 with cold atom recoil experiments .
as before , we make the following redefinitions in maggiore s proposal @xmath105 then , @xmath76 and @xmath77 satisfy the canonical commutation relations @xmath78=i \hbar \ , \delta_{ij}$ ] .
if we apply the redefinitions ( [ maggiorered ] ) to the hamiltonian eq .
( [ hamiltonian ] ) , we can get @xmath106 with this gup - corrected hamiltonian , the kinetic energy of @xmath0rb atoms is consequently modified into @xmath107 after a simple derivation , we can find a gup - modified relation @xmath108 together with eq .
( [ alpha ] ) , we finally get @xmath109 where terms involved in @xmath110 and @xmath111 have been omitted .
thus , the measurement ( [ hmeas ] ) , together with quantities in table i , sets an upper bound @xmath112 it is obvious that this bound on @xmath29 is much better than the one set by high - energy physics , which is very different from the case for the kmm proposal .
the reason is that @xmath29 is only associated with the factor @xmath92 in the calculation . without a further suppression due to the factor @xmath93 ,
the bound on @xmath29 is almost 13 orders better than the one on @xmath15 .
this feature makes maggiore s proposal easier to test than the kmm proposal .
we have discussed three proposals for the gup , and investigated how to use results from cold atom recoil experiments to constrain parameters in these proposals .
the main reason to take the @xmath0rb atom recoil experiment is that all the physical quantities involved can be measured with very high precision .
compared to bounds set from high - energy physics and other experiments , our bound on the kmm proposal is worse , which is mainly due to the suppression by the factor @xmath93 . in other words ,
cold atom recoil experiments are not suitable for studying effects of the kmm proposal .
things are totally different for the adv proposal and maggiore s proposal .
we get the second best bound on the adv proposal , and the best bound on maggiore s proposal . the high preciseness of the @xmath0rb atom recoil experiment plays the crucial role in giving these impressive bounds . however , these bounds are still many orders higher than theoretical predictions .
thus , to test various theories of quantum gravity , we have to resort to other possible atomic physics experiments .
this work was supported by the national key research program of china under grant no .
2016yfa0302002 , the national science foundation of china under grants no .
11227803 and no . 91536221 , and by funds from the chinese academy of sciences .
99 g. amelino - camelia , quantum - spacetime phenomenology , living rev . relativity , * 16 * , 5 ( 2013 ) .
m. maggiore , generalized uncertainty principle in quantum gravity , phys .
b * 304 * , 65 ( 1993 ) .
d. j. gross and p. f. mende , the high - energy behavior of string scattering amplitudes , phys .
b * 197 * , 129 ( 1987 ) ; d. j. gross and p. f. mende , string theory beyond the planck scale , nucl . phys .
b * 303 * , 407 ( 1988 ) .
d. amati , m. ciafaloni and g. veneziano , can space - time be probed below the string size , phys .
b * 216 * , 41 ( 1989 ) .
k. konishi , g. paffuti , and p. provero , minimum physical length and the generalized uncertainty principle in string theory , phys .
b * 234 * , 276 ( 1990 ) . i. elmashad , a. f. ali , l. i. abou - salem , j. u. nabi and a. tawfik , quantum gravity effect on the quark - gluon plasma , sop trans . on theor . phys . , * 1 * , 1 ( 2014 ) .
y. sabri and k. nouicer , phase transitions of a gup - corrected schwarzschild black hole within isothermal cavities , class .
* 29 * , 215015 ( 2012 ) .
a. tawfik , h. magdy and a. farag ali , effects of quantum gravity on the inflationary parameters and thermodynamics of the early universe , gen .
. grav . * 45 * , 1227 ( 2013 ) . v. m. tkachuk , deformed heisenberg algebra with minimal length and equivalence principle , phys .
a * 86 * , 062112 ( 2012 ) .
s. ghosh , quantum gravity effects in geodesic motion and predictions of equivalence principle violation , class .
. grav . * 31 * , 025025 ( 2014 ) .
s. das and e. c. vagenas , universality of quantum gravity corrections , phys .
* 101 * , 221301 ( 2008 ) .
a. f. ali , s. das and e. c. vagenas , proposal for testing quantum gravity in the lab , phys .
d * 84 * , 044013 ( 2011 ) .
i. pikovski , m. r. vanner , m. aspelmeyer , m. s. kim and c. brukner , probing planck - scale physics with quantum optics , nature phys .
* 8 * , 393 ( 2012 ) .
a. tawfik and a. m. diab , generalized uncertainty principle : approaches and applications , int . j. mod .
d * 23 * , 1430025 ( 2014 ) .
g. rosi , f. sorrentino , l. cacciapuoti , m. prevedelli and g. m. tino , precision measurement of the newtonian gravitational constant using cold atoms , nature * 510 * , 518 ( 2014 ) .
p. j. mohr , b. n. taylor and d. b. newell , codata recommended values of the fundamental physical constants : 2010 , rev .
phys . * 84 * , 1528 ( 2012 ) .
l. zhou , s. long , b. tang , x. chen , f. gao , w. peng , w. duan , j. zhong , z. xiong , j. wang , y. zhang and m. zhan , test of equivalence principle at @xmath9 level by a dual - species double - diffraction raman atom interferometer , phys .
. lett . * 115 * , 013004 ( 2015 ) .
g. ferrari , n. poli , f. sorrentino and g. m. tino , long - lived bloch oscillations with bosonic sr atoms and application to gravity measurement at the micrometer scale , phys .
lett . * 97 * , 060402 ( 2006 ) .
r. bouchendira , p. clad@xmath28 , s. guellati - kh@xmath28lifa , f. nez and f. biraben , new determination of the fine structure constant and test of the quantum electrodynamics , phys .
* 106 * , 080801 ( 2011 ) .
t. aoyama , m. hayakawa , t. kinoshita and m. nio , tenth - order qed contribution to the electron g-2 and an improved value of the fine structure sonstant , phys .
lett . * 109 * , 111807 ( 2012 ) .
a. d. cronin , j. schmiedmayer , and d.e .
pritchard , optics and interferometry with atoms and molecules , rev .
phys . * 81 * , 1051 ( 2009 ) .
( and references therein ) g. amelino - camelia , c. l@xmath113mmerzahl , f. mercati and g. m. tino , constraining the energy - momentum dispersion relation with planck - scale sensitivity using cold atoms , phys . rev
* 103 * , 171302 ( 2009 ) .
a. wicht , j. m hensley , e. sarajlic and s. chu , a preliminary measurement of the fine structure constant based on atom interferometry , phys .
t102 , 82 ( 2002 ) .
a. kempf , g. mangano and r. b. mann , hilbert space representation of the minimal length uncertainty relation , phys .
d * 52 * , 1108 ( 1995 ) .
a. f. ali , s. das and e. c. vagenas , discreteness of space from the generalized uncertainty principle , phys .
b * 678 * , 497 ( 2009 ) .
g. amelino - camelia , doubly - special relativity : first results and key open problems , int
d * 11 * , 1643 ( 2002 ) .
m. maggiore , quantum groups , gravity , and the generalized uncertainty principle , phys .
d * 49 * , 5182 ( 1994 ) .
p. clad@xmath28 , e. de mirandes , m. cadoret , s. guellati - kh@xmath28lifa , c. schwob , f. nez , l. julien and f. biraben , precise measurement of h/@xmath114 using bloch oscillations in a vertical optical lattice : determination of the fine - structure constant , phys .
a * 74 * , 052109 ( 2006 ) .
b. j. mount , m. redshaw and e. g. myers , atomic masses of @xmath115,@xmath116,@xmath117,@xmath118 , and @xmath119 , phys .
a * 82 * , 042513 ( 2010 ) . | various theories of quantum gravity predict the existence of a minimum length scale , which implies the planck - scale modifications of the heisenberg uncertainty principle to a so - called generalized uncertainty principle ( gup ) .
previous studies of the gup focused on its implications for high - energy physics , cosmology , and astrophysics . here , the application of the gup to low - energy quantum systems , and particularly cold atoms , is studied .
results from the @xmath0rb atom recoil experiment are used to set upper bounds on parameters in three different gup proposals . a @xmath1-level bound on the ali - das - vagenas proposal
is found , which is the second best bound so far .
a @xmath2-level bound on maggiore s proposal is obtained , which turns out to be the best available bound on it . 0.5 cm |
the seminal work of loss and divincenzo@xcite introduced the notion of using individual electrons trapped in gate - defined quantum dots to encode quantum information , an idea which has since burgeoned into a family of promising architectures for quantum computing @xcite .
an early theoretical realisation was that a single qubit encoded in three electron spins could be universally controlled using exchange couplings alone @xcite , which removes any requirement for individually addressable electron spin resonance or magnetic field gradients .
crucially , since exchange couplings in semiconductor experiments are controllable using gate voltages , this allows all qubit operations to be performed electronically ; an attractive feature in experimental implementations .
single - qubit operations for the so - called `` exchange - only '' qubit have been experimentally demonstrated @xcite .
pulse sequences are known for single - qubit gates that simultaneously correct leakage errors and other sources of noise @xcite . proposed two - qubit gates for the exchange - only qubit either involve capacitive coupling @xcite or exchange - coupling @xcite . since exchange coupling is usually much larger than capacitive coupling , exchange gates are usually faster , but come at the cost of requiring complicated pulse sequences in order to echo away the unwanted spin - flip transitions that occur as a side - effect and cause leakage errors .
the first such pulse sequence @xcite that effected a cnot required 19 exchange pulses in 13 timesteps , and was found using a numerical search .
since then , robust numerical searches have found improved pulse sequences that are robust against more sources of decoherence and/or reduce the number of gate operations @xcite . a recent alternative to the `` exchange - only '' qubit is the `` resonant - exchange qubit '' @xcite , which encodes qubits in the interaction picture with respect to significant exchange coupling between the three dots .
universal single qubit operations are effected using rf gate pulses to the electrodes controlling the exchange couplings .
this qubit has been shown both theoretically @xcite and experimentally @xcite to have several improved properties , including first - order insensitivity to charge fluctuations and reduced leakage error due to nuclear field fluctuations @xcite .
in addition , taylor et al . @xcite have shown that you can perform two - qubit gates between these qubits using charge dipole interactions , while we have suggested an alternative of using simple exchange pulses between nearby qubits @xcite . in this earlier work @xcite , we showed that a two - qubit cphase gate can be implemented using a single exchange pulse between the constituent quantum dots of neighbouring qubits ( shown schematically in figure [ fig : physical - configurations ] ) .
rapid high - fidelity gate operation is in principle made possible by energetically suppressing the spin - flip transitions that lead to leakage .
this method of effecting a two - qubit gate contrasted with the more conventional approach of long and complicated pulse sequences , and is similar to our earlier proposal for singlet - triplet qubits @xcite .
it is the purpose of this work to extend our previous results by considering higher - order analysis , adiabatic pulse profiles , the effect of noise , and the performance of our gate in physically motivated simulations that include noise . in section [ sec : resonant - exchange - qubits ] we briefly review single - qubit resonant - exchange qubit operations , in section [ sec : our - two - qubit ] we review our two - qubit gates , in section [ sec : gate - characterisation ] we formally characterise our two - qubit gate in the butterfly geometry , in section [ sec : alternative - geometries ] we briefly consider other geometries , and in section [ sec : conclusions ] we conclude . in this work
we set @xmath0 , meaning that energies are interchangeable with angular frequencies .
a resonant exchange qubit is a triple - dot system operating deep in the ( 1,1,1 ) charge state ( each dot almost surely confines a single electron ) .
a large magnetic field is applied along the @xmath1-axis , zeeman splitting the @xmath2 spin states according to the z - projection of their total spins : @xmath3 , @xmath4 is the effective zeeman splitting of an electron subject to the global magnetic field .
note that we have absorbed the g - factor ( up to its sign ) into our definition of @xmath4 .
the logical states of the qubit are @xmath5 and @xmath6 , which both have total spin @xmath7 with z - projection of @xmath8 .
the remainder of the eight - dimensional hilbert space describes non - logical states , the spanning eigenstates of which are completed for the energy eigenbasis in table [ tab : qubit - states ] .
these qubits are operated with large intra - qubit exchange couplings @xmath9 and @xmath10 , with oscillatory modulations around @xmath11 providing single qubit control @xcite .
this is depicted and described in more detail in figure [ fig : single - qubit - control ] .
and @xmath10 couplings , states rotate about the same axes as those of the rotations indicated in green and red respectively . if both @xmath9 and @xmath10 are equal to @xmath12 , states rotate about the @xmath1 axis as indicated by the emboldened blue rotation . pulses involving different @xmath9 and @xmath10 couplings allow arbitrary rotations , and hence provide universal single qubit control . as these intra - qubit couplings are controllable using gate voltages , this allows for universal electronic control of qubits.[fig : single - qubit - control ] ] intra - qubit exchange couplings are in turn controlled by detuning the voltages defining the quantum dots ( which we parametrise as @xmath13 ) .
this allows for complete electronic control of qubits , which potentially simplifies experimental implementation . in several gaas singlet - triplet qubit experiments
@xcite , an exponential ansatz @xmath14 has been found to be a good phenomenological fit to experimental data over a wide range of interesting values of @xmath13 , and so we adopt it in this work .
.energy eigenstates of a single resonant exchange qubit , sorted by @xmath15 .
@xmath16 is the total angular momentum quantum number of the three electron spins and @xmath15 is the z - component of the total angular momentum .
@xmath17 is the energy level splitting due to exchange coupling , and @xmath4 is the splitting of an electron subject to the large transverse global field .
each eigenstate with a negative subscript has their constituent spins flipped relative to the corresponding unsubscripted state.[tab : qubit - states ] [ cols="^,^,^,^,^",options="header " , ]
in this section we describe how the effective hamiltonian of equations 1 and 2 in the main text is lifted from perturbation theory , the procedure for which follows standard practice @xcite .
the main idea is that , in the limit that inter - qubit coupling @xmath18 is small compared to intra - qubit coupling @xmath12 , the evolution of the logical subspace in which we are interested should be well - approximated by low - order terms in a perturbation expansion around @xmath19 ; from which considerable insight might be gained into qubit dynamics . to generate the effective hamiltonian for a given geometry , we perturb the hamiltonian describing the two decoupled resonant exchange qubits with the hamiltonian describing the two - qubit coupling of that geometry , to first order in a rayleigh - schroedinger perturbation expansion .
we then construct the effective hamiltonian using the resulting energies @xmath20 and eigenvectors @xmath21 that adiabatically map to the logical states using @xmath22 .
this algorithm explicitly disregards any leakage operations , but captures the dominant dynamics on the logical subspace .
the result is a diagonal hamiltonian , except where the original logical states were degenerate . for our two resonant exchange qubit system ,
in which the @xmath23 and @xmath24 states are degenerate ( shown in figure 3 of the main text ) , this results in a block - diagonal hamiltonian when written in the basis @xmath25 .
we explicitly compute the effective hamiltonians for the geometries shown in figure 2 of the main text ; that is , the linear , butterfly and rectangular geometries respectively .
we also explicitly compute the coefficients of @xmath26 , @xmath27 , @xmath28 , and @xmath29 , which are then used to populate table ii of the main text .
@xmath30\ ] ] @xmath31\ ] ] @xmath32\ ] ] @xmath33\ ] ] @xmath34\ ] ] @xmath35\ ] ] note that to recover the provided coefficients in table ii of the main text , we assume that all inter - qubit couplings ( @xmath36 , @xmath37 and @xmath38 ) are equal to @xmath18 .
as computed using the `` hybrid '' method described in the text , which is our best approach for maximising the fidelity of gate operations .
@xmath39 is defined as in the text , and plotted here for @xmath40 . in ( b ) we plot the relative difference @xmath41 of the other methods described in the text , namely `` adiabatic '' ( green dashed ) and `` first order '' ( red dotted ) methods , to the `` hybrid '' estimate ( blue solid ) .
the plots share the same x axis , which is the ratio of intra - qubit coupling to inter - qubit coupling @xmath42 .
note that the first order approximation consistently underestimates the phase accrual rate , and so overestimates the gate time .
while an adiabatic phase estimation does better , it ends up underestimating the phase due to the importance of evolution off the logical subspace for large enough @xmath18.[fig : timing - errors ] ] as mentioned in section iva , over- or under - estimating the ideal gate time @xmath43 leads to over- or under - accrual of two - qubit phase , resulting in poor gate fidelities .
in particular , using a first order approximation neglects higher order terms that give rise to convex non - linearity in @xmath44 , and so consistently exaggerates gate times . in this section , we describe the method by which we obtain accurate timing estimates for use in the simulations of the main text .
there are two methods that one might employ to find an estimate of @xmath43 that includes high order perturbations .
one can make the assumption that evolution is perfectly adiabatic , solve for the eigenvalues at different values of inter - qubit coupling @xmath18 , and thus infer the two - qubit phase accumulation rate @xmath45 ; or , alternatively , one can locally maximise the gate fidelity over gate times .
the first method works well for small values of @xmath42 ( where the adiabatic approximation make sense ) , and the second method works well for larger values of @xmath42 ( where gate times are shorter , and numerical integration has less time to accumulate error ) .
we therefore use a hybrid approach that works well across all values of @xmath42 : for @xmath46 , we use the adiabatic approach , and for @xmath47 we use the adiabatic approach to seed a numerical optimisation of gate fidelity over gate times . as noted in the main text , the non - linearity of @xmath44 means that we must repeat this estimation process for each pulse shape considered . as the dynamics of our ( noiseless ) two - qubit system is determined by the ratio of @xmath42 , given a particular pulse shape , we need only optimise over the ratio of the two physical degrees of freedom , rather than both .
we therefore proxy the optimisation of @xmath48 for any given pulse shape by optimisations over the single parameter function @xmath49 . in practice
we fix a value of @xmath50 , and vary @xmath18 .
@xmath43 can be recovered from @xmath51 using : @xmath52 the gate time for maximum fidelity @xmath53 for the butterfly configuration using a sinusoidal adiabatic pulse profile is shown in figure [ fig : timing - errors ] , along with the relative error of the first order and adiabatic estimations .
we note that the first order approximation consistently underestimates the phase accrual rate , and so overestimates the gate time .
while an adiabatic phase estimation does better at first , it ends up underestimating the phase due to the importance of evolution off the logical subspace for large enough @xmath18 .
simulations involving the overhauser field modelled as described in the main text requires averaging over a multi - variate normal distribution in 6 variables ( 11 if dc charge noise is also considered ) . while a naive monte - carlo simulation that samples the local overhauser field contribution for each dot will converge eventually , we speed up the convergence by instead sampling preferentially from linear combinations of local overhauser contributions that appear at lower order in perturbation theory . inspired by perturbation theory , we form a spanning basis for the magnetic fields : @xmath54 where @xmath55 ( with @xmath56 $ ] ) is the z - component of the overhauser field at dot @xmath57 . we learn from perturbation theory that @xmath58 and @xmath59 first contribute at zeroth order in perturbation theory , @xmath60 and @xmath61 at first order , and @xmath62 at fourth order . as @xmath63 is a global field , it does not contribute .
the higher the order at which the terms contribute , the more significant that term is to the dynamics of the qubit system .
it makes sense , therefore , to sample lower order terms more often .
technically , one should weight each term roughly as @xmath64 , where @xmath65 is the order at which the term appears in perturbation theory .
we found it simpler , however , to use a conservative ratio of @xmath66 between terms of different order .
that is , for each 100 simulations , we sample @xmath58 and @xmath59 100 times , @xmath60 and @xmath61 20 times , and @xmath62 4 times .
as the terms of the new basis are linear compositions of the @xmath55 terms , the random variable associated with each term will differ from the underlying local field fluctuations .
in particular , if each @xmath67 , then the relevant terms of new basis will be sampled from the distributions : @xmath68
perturbation theory provides the insight that beyond first order the overhauser field cross - couples with charge noise in @xmath18 at second order and above .
this allows the overhauser field to contribute non - trivially to two - qubit phases .
for example , the two - qubit phase accumulation rate to second order for the butterfly configuration is given by : @xmath69 where @xmath70 , and where the coloured and starred line is the effect of cross - coupling .
the intuition is that as @xmath42 and the @xmath71 increase , the greater the two - qubit cross - coupling error . for the useful range of @xmath42 values , and
the experimentally inspired values of @xmath71 used in the main text , the effect of this cross - coupling is negligible . | resonant exchange qubits are a promising addition to the family of experimentally implemented encodings of single qubits using semiconductor quantum dots . we have shown previously that it ought to be straightforward to perform a cphase gate between two resonant exchange qubits with a single exchange pulse .
this approach uses energy gaps to suppress leakage rather than conventional pulse sequences . in this paper
we present analysis and simulations of our proposed two - qubit gate subject to charge and overhauser field noise at levels observed in current experiments .
our main result is that we expect implementations of our two - qubit gate to achieve high fidelities , with errors at the percent level and gate times comparable to single - qubit operations . as such
, exchange - coupled resonant exchange qubits remain an attractive approach for quantum computing . |
understanding the electronic nature of chemical bonds of @xmath7-elements is an issue which has been attracting wide attention of investigators working in diverse areas of research .
the main question is the involvement of 4@xmath7 states in the formation of chemical bonds in molecules and in solids . in general
, one can speak of two extreme scenarios .
that is , either @xmath8electrons participate in chemical bonding which is the case for 5@xmath7-states in the u@xmath0 molecule @xcite , or they stay essentially localized at each site forming core - like states like in lanthanides .
this phenomenon known as the localization - delocalization paradigm in solid - state physics @xcite is the subject of intense theoretical considerations and discussion . in various compounds cerium
is believed to display both types of 4@xmath7 behavior which makes it the key element in these studies . indeed , in solids ce and its compounds exhibit a broad range of unusual electronic , magnetic and structural properties called correlation effects @xcite . while much effort has been invested in understanding of the complex electronic behavior in sophisticated cerium compounds ,
very little attention has been given to the simplest cerium chemical bond @xcite , which is realized in ce@xmath0 .
meanwhile , owing to the progress of the spectroscopy of small metal clusters @xcite precise experimental data on the cerium dimer @xcite have been reported recently , followed by first reliable many - electron calculations @xcite .
cao and dolg @xcite have performed a series of calculations of ce@xmath0 using the complete active space self - consistent field ( casscf ) method , multireference configuration interaction ( mrci ) method and coupled cluster approach [ ccsd(t ) ] and concluded that its ground state superconfiguration " is @xmath9@xmath10 .
six low - lying molecular terms ( @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 and @xmath16 ) are virtually degenerate " so that the exact refinement of the ground state is very difficult .
the authors then restricted themselves to the @xmath11 and @xmath17 ground states for which in the scalar relativistic approximation they obtained the following spectroscopic constants : @xmath18 , @xmath19 ev and @xmath20 @xmath21 .
the low - lying excitations were attributed to the core - like properties of @xmath1 states . in the spin - orbit calculation
the authors obtained the @xmath22_g^+$ ] ground state followed by two low - lying levels @xmath23
( @xmath24 1 @xmath21 ) and @xmath25 ( @xmath24 2 @xmath21 ) .
the nature of the ce@xmath0 bonding and the role of the 4@xmath7-states were also considered in casscf studies by roos _
et al . _ in @xcite . as in @xcite
it was found that the triple bond molecular configuration @xmath26 couples in the @xmath11 molecular term .
two remaining nonbonding and localized 4@xmath7 electrons are in @xmath27-state ( @xmath28 ) .
these two 4@xmath7 electrons then can form the configurations @xmath29 , @xmath30 , and @xmath31 , which result in molecular terms with angular momenta @xmath32 and 6 .
the six calculated low - lying levels again are almost degenerate having approximately the same spectroscopic parameters ( without the spin - orbit interaction ) : @xmath33 , @xmath34 @xmath21 , @xmath35 ev ( the caspt2 level of theory and a basis set of vqzp quality ) .
the spin - orbit coupling has only small effect on the computed properties , in particular it reduces @xmath36 to 2.51 ev in good agreement with the experimental estimate @xmath37 ev [ from the third law ] .
calculations of roos _ et al . _
give a longer @xmath38 and a softer @xmath39 in comparison with the values of cao and dolg @xcite .
both theoretical estimates of @xmath39 @xcite are smaller than the experimental result @xcite : @xmath40 @xmath21 [ the force constant @xmath41 mdyne / ] .
thus , the standard well - known many - electron procedures @xcite correctly describe the ce@xmath0 chemical bonding but find six nearly degenerate low - lying molecular terms ( @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 and @xmath16 ) instead of a well defined ground state .
the difficulty lies in the smallness of the splitting energies of these excitations requiring a special treatment . in the following
we formulate a simplified valence bond ( vb ) model accounting for the 4@xmath7 electrons in the cerium dimer , find the important interactions and obtain the fine " energy spectrum of the electron excitations .
our approach is based on the picture of 4@xmath7 - 4@xmath7 interactions as rightly found in @xcite , but requires a more deep insight in the couplings which experience the 4@xmath7 electrons in the molecule .
we start with the description of the triple bond of ce@xmath0 .
the ground state electronic configuration of the cerium atom is @xmath42 with the @xmath43 ground term .
this implies that the ce atom is a rare case of the first hund rule violation @xcite .
the nature of such behavior lies in quartic fermi hole " which effectively separates the valence electrons and decreases their coulomb repulsion @xcite . already at this stage of the investigation
it is clear that the 4@xmath7 electron is not independent , through the coulomb repulsion it is strongly coupled with the valence 5@xmath44 and 6@xmath45 electrons on the same cerium site .
however , the @xmath42 configuration with the closed @xmath46shell is not favorable for chemical bonding .
therefore the electronic atomic state of ce is first transformed to the @xmath47 configuration with the @xmath48 term situated at 0.299 ev above the @xmath43 ground term @xcite . in the @xmath47 configuration
both 5@xmath44 and 6@xmath45 shells become open and form the triple bond with six valence electrons : @xmath10 .
the @xmath11 bonding state of 5@xmath44 and 6@xmath45 electrons ( without the 4@xmath7 states ) is well understood @xcite , in particular it occurs in la@xmath0 where the molecular characteristics ( @xmath49 ev , @xmath41 mdyne / @xcite , the @xmath50 initial and @xmath51 final [ at 0.33 ev ] configuration ) are very close to those in ce@xmath0 .
although the triple bond is strong , the total energy gain is weakened by 0.6 ev because of the necessity to promote one 6@xmath45-electron to the 5@xmath44-shell for both cerium atoms .
we consider the first cerium atom at ( 0,0,0 ) and the second at ( 0,0,@xmath38 ) ( so that the @xmath52-axis is the molecular axis ) and find the valence density distribution of the four 5@xmath44 and two 6@xmath45 valence electrons .
the @xmath11 @xmath53 ground state implies an axially symmetric electron density .
the conclusion can also be proved by considering the density distributions of two @xmath45 and four @xmath44-electrons in the @xmath10 configuration .
now to the @xmath54 and @xmath55 valence states we add two @xmath1-electrons .
the main assumption supported by @xcite is that to a first approximation the 4@xmath7 electrons do not contribute to the chemical bonding and remain localized .
then it follows that each 4@xmath7 electron should be either in the @xmath56 or in the @xmath57 angular orbital state where @xmath58 .
( here we deal with the real spherical harmonics @xmath59 ( @xmath60 for cosine- and @xmath61 for sine - like azimuthal ( @xmath62- ) dependence ) defined in ref .
@xcite . ) that is because in both these states the coulomb intrasite repulsion between the @xmath1 electron and the bonding @xmath54 and @xmath55 states at the cerium site is minimal ( see appendix a ) . indeed , in these @xmath28 @xmath8states the angular density dependence is @xmath63 with the maximum at @xmath64 , while in the binding @xmath54 states maximum lies at @xmath65 and in @xmath55 states at @xmath66 ( considering their overlap ) .
thus , the occupation of the @xmath28 @xmath8states ensures a certain space separation of the populated @xmath1 states from the binding @xmath54 and @xmath55 electrons , which effectively minimizes the intrasite @xmath67 and @xmath67 coulomb repulsions .
the situation is discussed in detail in appendix a , notice that the conclusion about the occupancy of the @xmath68 states is also explicitly supported by the calculations of roos et al .
@xcite and implicitly by cao and dolg @xcite .
thus , at each cerium site the @xmath1 electron space is limited to two @xmath28 orbitals , see fig .
[ fig1 ] : @xmath69 here @xmath70 is the radial part of the @xmath1 states
. since there is very little overlap between @xmath1 functions belonging to different sites , the valence bond ( vb or heitler - london ) @xcite approach should give the best description to the @xmath1 electronic problem . its basis set consists of the two electron @xmath1 states : @xmath71 where the indices @xmath72 and @xmath73 refer to two orbital sites at sites @xmath74 and @xmath75 , i.e. @xmath76 or @xmath45 ( orbital functions defined in eq .
( [ f1a],b ) and visualized in fig .
[ fig1 ] ) , @xmath77 is the spin projection number ( @xmath78 ) , @xmath79 is the corresponding spin function ( @xmath80 or @xmath81 ) .
the two electron wave function @xmath82=@xmath83 is a product of the orbital part @xmath84 ( @xmath85 ) , which can be labeled as @xmath86 and the spin function @xmath87 ( @xmath88 ) , @xmath89 ( here we omit cerium site indices @xmath74 and @xmath75 , that is @xmath90 etc . ) therefore , in total there are 16 two electron basis states @xmath91 ( @xmath921 - 16 ) , which are then antisymmetrized in respect to the electron permutation . in atomic units
the corresponding valence bond hamiltonian @xcite is given by @xmath93 where @xmath94 is the spin - orbit coupling considered in detail below ( sec .
[ subsec - so ] ) , while @xmath95 here @xmath96 is an effective @xmath1 attraction to the nucleus @xmath74 ( or @xmath75 ) screened by all core electrons .
it is well known that the evaluation of the matrix elements of @xmath97 in the space of the functions ( [ f2 ] ) leads to the following equation : @xmath98 here @xmath99 $ ] is the background molecular energy , @xmath100 is the kronecker delta , while @xmath101 and @xmath102 are the direct and exchange matrix elements which we consider in sec .
[ subseca ] and sec .
[ subsecb ] , respectively . taking into account the orbital degeneracy of 4@xmath7 states the matrix elements of @xmath103 ( 16@xmath10416 ) where @xmath105 , @xmath102 or @xmath94 are conveniently written in the spin block form , @xmath106 where the indices @xmath107 refer to the spin part , eq .
( [ f2b ] ) . as usual , @xmath108 etc .
each of the blocks in eq .
( [ f3a ] ) can be further represented as a 4@xmath1044 orbital matrix of the following structure : @xmath109 here the matrix elements are evaluated between the two @xmath8orbital sets [ @xmath110 , eq .
( [ f2a ] ) ] with the fixed spin indices [ @xmath107 , eq .
( [ f2b ] ) ] . in our basis with real spherical harmonics , @xmath111 , @xmath112 , @xmath113
are real , while @xmath114 and @xmath115 are imaginary values describing the interaction with an external magnetic field or the spin - orbit coupling .
the specific form of the matrix elements is considered below . here
we consider the matrix of the direct coulomb repulsion : @xmath117 where @xmath118 and the orbital indices @xmath119 belong to the composite index @xmath120 , while @xmath121 to @xmath122 , eq . ( [ f2 ] ) . from symmetry arguments
it can be shown ( appendix b ) that there are three types of the matrix elements , @xmath123 while @xmath124 here we use both standard and so called chemist s notations @xcite .
however , it turns out that not all parameters in ( [ f4 ] ) are independent . analyzing the azimuthal dependence of the integrands and the azimuthal selection rule " ( appendix b )
, one finds @xmath125 in the @xmath105 case in eq .
( [ f3a ] ) only the diagonal blocks @xmath126 are relevant ( @xmath127 for @xmath128 ) because the matrix elements ( [ h2 ] ) are independent of the spin indices .
furthermore , the @xmath126 blocks are identical , i.e. @xmath129 , eq . ( [ f3 ] ) .
the @xmath130 matrix is given by eq .
( [ f3 ] ) where @xmath131 , @xmath132 , @xmath133 and @xmath134 . for the matrix element of the @xmath116 exchange interaction
we obtain @xmath135 where the composite indices @xmath120 and @xmath122 refer to @xmath136 and @xmath137 as described earlier , eq .
( [ f2 ] ) , and @xmath138 finally , the coulomb exchange integral in ( [ f8a ] ) is given by [ compare with @xmath139 in eq .
( [ h2 ] ) ] @xmath140 if two first terms on the right hand side of ( [ f8a ] ) differ from zero then they are leading , while the third ( coulomb ) term is just a minor correction .
in our case there are two @xmath1 functions at each site , eq .
( [ f1a])-([f1b ] ) , but from symmetry it follows that in all overlap integrals @xmath141 there is only one nonzero value , @xmath142 while @xmath143 .
analogously , @xmath144 and @xmath145 .
it is convenient to introduce the notation @xmath146 for the leading term . for the exchange coulomb interaction @xmath147 , eq .
( [ f8c ] ) , one finds three types of integrals with nonzero values ( compare with eqs .
( [ f4 ] ) ) : @xmath148 furthermore , there is the same relation as for the direct matrix elements @xmath149 , @xmath150 and @xmath151 , eq . ( [ f5a ] ) ,
see appendix b : @xmath152 then , in the full 16@xmath10416 matrix @xmath153 , eq . ( [ f3a ] ) , the exchange coupling is effective in the blocks @xmath154 ( between states @xmath80 , @xmath80 ) , @xmath155 and @xmath156 ( between the states with @xmath80 , @xmath81 and @xmath81 , @xmath80 ) and @xmath157 ( between states @xmath81 , @xmath81 ) .
each of the blocks has the following ( orbital ) form : @xmath158 ( here as in ref .
@xcite we have taken into account the effect of the half canceling of the @xmath159-terms in the final expression due to the nonorthogonality of the @xmath7-functions . ) now in the framework of our model we consider the @xmath1 spin - orbit interaction : @xmath160 where @xmath161 k. this value corresponds to the @xmath1620.26 ev splitting between @xmath163 and @xmath164 one electron atomic @xmath1 states .
( the energies of the states were obtained by solving the spherically symmetric dirac equation for atomic cerium states in the framework of the self - consistent field dft - lda atomic model . ) in our model the @xmath1 quantum space is reduced to two orbital states with @xmath165 at each site , and @xmath94 transforms to @xmath166 therefore , the nonzero spin blocks of the full matrix are @xmath167 ( @xmath168 ) , eq .
( [ f3a ] ) , with the following general structure of each block ( see appendix c ) : @xmath169 in the @xmath170 block @xmath171 , @xmath172 , in the @xmath173 block @xmath174 , @xmath175 , in the @xmath176 block @xmath177 , @xmath178 and in @xmath179 block @xmath180 , @xmath181 .
the spin - orbit coupling , eq .
( [ f21 ] ) , splits the four single electron states at each cerium site in two doublets : the lowest level with @xmath182 and the energy @xmath183 has the eigenfunctions @xmath184 , @xmath185 .
the other level with @xmath186 and the energy @xmath187 has the functions @xmath188 , @xmath189 .
correspondingly , two 4@xmath7 electrons have the following energy spectrum : @xmath190 ( 4 degenerate states ) , @xmath191 ( 8 states ) , @xmath192 ( 4 states ) .
the inclusion of other interactions modifies the picture .
all interaction parameters of the model introduced in sec . [ sec : model ] were evaluated numerically with our valence bond program @xcite , table [ table1 ] .
there , we have used the overlapping cerium relativistic atomic @xmath1 functions ( the large @xmath163 component for @xmath70 ) calculated within the density functional theory ( dft ) in the local density approximation ( lda ) @xcite . .
calculated @xmath1 direct ( @xmath193 ) and exchange ( @xmath194 ) coulomb integrals , the overlap integral @xmath195 , the vb attraction @xmath196 and the @xmath1 exchange parameter @xmath159 for various ce - ce distances @xmath197 ( in ) .
all energy parameters are in kelvin ( k ) .
+ @xmath198 ce@xmath0 equilibrium distance for @xmath11 , @xmath14 in ref .
@xcite , + @xmath199 ce@xmath0 equilibrium distance for @xmath11 , @xmath15 , @xmath14 , @xmath200 in ref .
@xcite , + @xmath201 ce@xmath0 equilibrium distance for @xmath12 , @xmath13 in ref .
+ @xmath2023.43 and 3.65 correspond to the nearest neighbor distance in @xmath80-ce and @xmath203-ce @xcite , respectively .
[ table1 ] [ cols="^,^,^,^,^,^,^",options="header " , ] thus , we conclude that the @xmath204 coulomb repulsion favors the lowering of the @xmath28 @xmath8states . however , there are two other interactions which act in the opposite direction : the screened attraction to the other nucleus and the @xmath204 exchange interaction .
the other nucleus attraction leads to an axially symmetric coulomb potential which has the form of ( [ a3 ] ) , but with an interaction constant of the opposite sign .
the @xmath204 exchange interaction is more complex . in terms of the intrasite coulomb multipole repulsion it corresponds to the odd ( @xmath205 ) transition density contributions , and the dipolar ( @xmath206 )
contribution is rather strong @xcite .
however , this exchange interaction leads to the high spin ( @xmath207 ) ground state ( the @xmath48 term of the @xmath208 configuration ) of the cerium atom , which in the real molecule is effectively suppressed by the triple bond formation with a zero spin state of the binding @xmath54 and @xmath55 states .
therefore , the bond formation mechanism diminishes the @xmath209 exchange interaction , while the @xmath209 coulomb repulsion ( [ a3 ] ) remains and lowers the @xmath28 @xmath8states in respect to the others @xcite .
the relevant coulomb integrals , eq . ( [ f4 ] ) , can be simplified by considering their azimuthal parts , i.e. the azimuthal dependence of the corresponding @xmath210 functions standing on the right and their counterparts @xmath211 on the left from the coulomb interaction @xmath212 , eq .
( [ h2 ] ) . in the following
we define @xmath213 and @xmath214 .
the important observation is that the coulomb integral differs from zero only if the azimuthal density dependencies on the left and on the right side have the same functional form .
this conclusion can be derived by various ways , in particular it follows from the multipole expansion of the coulomb interaction @xmath215 here the angular functions @xmath216 and @xmath217 have the same azimuthal dependence ( @xmath218 ) which is a trivial conclusion if they refer to the same site : @xmath219 . in case of two different sites ( aligned along the @xmath220axis )
one can expand a spherical harmonic centered at one site in terms of the spherical harmonics centered at the other site @xcite .
it turns out that the procedure conserves the azimuthal dependence of the spherical harmonics translated along the @xmath220axis @xcite , which proves our initial statement . in the following in deriving relations between the coulomb integrals , eq .
( [ f4 ] ) , we use chemist s notations for the matrix elements @xcite .
further , we explicitly single out the azimuthal dependence @xmath221 out of the density functions @xmath222 with @xmath223 , eq . ( [ f1a],b ) .
we then arrive at three different dependencies : @xmath224 from the azimuthal selection rule it follows that @xmath225 , eq .
( [ f4 ] ) and ( [ f5b ] ) , and @xmath226 . in order to prove the relation ( [ f5a ] ) we consider the structure of the general coulomb integral , @xmath227 where @xmath228 stands for @xmath229 , while @xmath230 is the azimuthal integral with two identical @xmath62-dependencies @xmath231 and @xmath232 .
for example , @xmath233 with @xmath234 stand for the azimuthal integral between @xmath235 and @xmath235 or between @xmath236 and @xmath236 : @xmath237 each term on the right hand side of ( [ b3 ] ) has the following form : @xmath238 ^ 2 \nonumber \\
r_f^2(r_a ) r_f^2(r_b)\ , i_{a}(m_a|m_b ) .
\quad \quad \label{b4}\end{aligned}\ ] ] here @xmath239 is the associated legendre polynomial @xcite , @xmath58 , @xmath70 is the radial dependence of the @xmath1 functions . taking into account three relevant azimuthal dependencies ( [ b2a])-([b2c ] )
we get @xmath240 from eqs .
( [ b5 ] ) we immediately obtain @xmath241 or @xmath242 , eq .
( [ f5a ] ) .
for the exchange charge @xmath243 ( @xmath244 ) which corresponds to the @xmath1 electron transition from the site @xmath74 to the site @xmath75 we again have the three azimuthal dependencies , eqs .
( [ b2a])-([b2c ] ) . therefore ,
proceeding as before albeit with a modified expressions for the integrals we obtain the analogous relation @xmath245 or @xmath246 , eq .
( [ f12 ] ) .
as before , the result is based solely on the azimuthal selection rule .
the action of the @xmath220component of the orbital momentum on the real spherical harmonics @xmath247 , @xmath248 @xcite ( the @xmath52-axis coincides with the molecular axis ) is given by @xmath249 the matrix elements of @xmath250 in the space of two electron @xmath1 states ( eqs .
( [ f2a ] ) and ( [ f2b ] ) ) can be obtained from these relations .
explicitly , the matrix of @xmath251 has the nonzero blocks @xmath252 ( @xmath168 ) [ see eq .
( [ f3a ] ) ] of the form @xmath253 where @xmath254 is the bohr magneton .
further , the 16@xmath10416 spin matrix @xmath255 , eq .
( [ f3a ] ) , has two nonzero diagonal blocks @xmath256 and @xmath257 of the form @xmath258 where the plus sign refers to the @xmath259 block ( @xmath260 ) and the minus sign to the @xmath261 block ( @xmath262 ) , while @xmath263 stands for the unit 4@xmath1044 matrix .
the molecular interaction with an external magnetic field @xmath103 is given by @xmath264 with the molecular magnetic moment @xmath265 for the magnetic field @xmath103 applied along the @xmath220axis we obtain @xmath266 where @xmath267 .
the relations ( [ c2 ] ) , ( [ c3 ] ) and ( [ c6 ] ) are used to calculate the @xmath52-component of the orbital , spin and full momentum for each two - electron level .
for example , by means of eq .
( [ c2 ] ) we can find the orbital quantum number required for the characterization of the degenerate states , which are considered in sec .
[ subseca ] .
there we have discussed the three energy levels given by eq .
( [ f6a])-([f6c ] ) .
the corresponding eigenvectors are in the presence of the external magnetic field the hamiltonian @xmath269 of the direct coulomb repulsion ( see sec . [ subseca ] ) is modified : @xmath270 where the last term is the orbital polarization given by @xmath271 ( the first term @xmath272 is the binding energy of ce@xmath0 which is not important for our discussion ) . from ( [ c7 ] )
we obtain four eigenvalues ( @xmath85 ) @xmath273 and the corresponding eigenvectors : @xmath274 , @xmath275 , @xmath276 , and @xmath277 .
notice , that at zero magnetic field ( @xmath278 ) @xmath279 as in ( [ f6b ] ) . on the other hand , from ( [ c8 ] )
it is clear that two components correspond to @xmath280 and therefore in this doublet @xmath281 .
in other two non - degenerate states @xmath282 ( i.e. they are @xmath283-states ) .
l. gagliardi , b.o .
roos , nature * 433 * , 848 ( 2005 ) .
albers , nature * 410 * , 759 ( 2001 ) .
p. fulde , _ electron correlations in molecules and solids _
( springer heidelberg , 1995 ) .
x. cao and m. dolg , mol .
* 101 * , 1967 ( 2003 ) .
roos , r. lindh , p .- a .
malmqvist , v. veryazov , and p .- o .
widmark , j. phys .
a * 112 * , 11431 ( 2008 ) .
x. shen , l. fang , x. chen , and j.r .
lombardi , j. chem .
phys . * 113 * , 2233 ( 2000 ) .
j.r . lombardi and b. davis ,
chem . rev . * 102 * , 2431 ( 2002 ) .
morgan iii and w. kutzelnigg , j. phys .
chem . * 97 * , 2425 ( 1993 ) .
e. f. worden , r. w. solarz , j. a. paisner , and j. g. conway , j. opt .
b * 68 * , 52 ( 1978 ) ; w. c. martin , r. zalubas , and l. hagan , _ atomic energy levels - the rare earth elements _ , natl . stand . ref .
* 60 * ( 1978 ) .
bradley and a.p .
cracknell , _ the mathematical theory of symmetry in solids _ , ( clarendon , oxford , 1972 ) .
w. heitler and f. london , z. physik * 44 * , 455 ( 1927 ) . ,
cooper ( editor ) , theoretical and computational chemistry , vol . 10
( elsevier , amsterdam , 2002 ) .
a.v . nikolaev and k.h .
michel , j. chem .
phys . * 122 * , 064310 ( 2005 ) .
parr , w. yang , _ density - functional theory of atoms and molecules _ , ( new york , oxford university press , 1989 ) .
koskenmaki and k.a .
gschneidner , jr . , _
handbook on the physics and chemistry of rare earths _
gschneidner , jr . , and l. eyring ( amsterdam : north - holland , 1978 ) , p. 337 .
a. szabo and n.s .
ostlund , _ modern quantum chemistry _ , ( mcgraw - hill , dover , 1989 ) .
landau and e.m .
lifshitz , _ quantum mechanics - non - relativistic theory _ , v. 3 , ( pergamon press , oxford , 1965 ) , chap .
tsvyashchenko , a.v .
nikolaev , a.i .
velichkov , a.v .
salamatin , l.n .
fomicheva , g.k .
ryasny , a.a .
sorokin , o.i .
kochetov , m. budzynski , and k.h .
michel , phys .
b * 82 * , 092102 ( 2010 ) .
a.v . nikolaev and k.h .
michel , phys .
b * 66 * , 054103 ( 2002 ) . | the electronic structure of ce@xmath0 is studied in a valence bond model with two @xmath1 electrons localized at two cerium sites .
it is shown that the low - lying energy spectrum of the simplest cerium chemical bond is determined by peculiarities of the occupied @xmath1 states .
the model allows for an analytical solution which is discussed along with the numerical analysis .
the energy spectrum is a result of the interplay between the @xmath1 valence bond exchange , the @xmath1 coulomb repulsion and the spin - orbit coupling .
the calculated ground state is the even @xmath2 level , the lowest excitations situated at @xmath3 k are the odd @xmath2 state and the @xmath4 doublet @xmath5 .
the calculated magnetic susceptibility displays different behavior at high and low temperatures . in the absence of the spin - orbit coupling the ground state
is the @xmath6 triplet .
the results are compared with other many - electron calculations and experimental data . |
dust plagues the recovery from observations of the intrinsic ( i.e. , unaffected by the dust ) total and two - dimensional ( i.e. , 2-d ) photometric parameters of the dusty stellar systems .
its effects vary as a function of the wavelength of the photons that are probed by a given broad - band filter ( see calzetti 2001 for a recent review ) .
radiative transfer calculations for simplistic dust / stars configurations , simulating dusty late - type galaxies , offer equivocal results for the attenuation by internal dust of the stellar light produced in these systems ( disney , davies , & phillips 1989 ) .
they also frequently do not describe the observations over a large spectral range ( kuchinski et al .
1998 ) . furthermore , _
realistic radiative transfer simulations have to include scattering from dust grains _
( bruzual , magris , & calvet 1988 ; witt , thronson , & capuano 1992 ; emsellem 1995 ) , a phenomenon that is not negligible given the high albedo observed e.g. for galactic dust ( gordon , calzetti , & witt 1997 ; gordon 2004 ) . to make computations even more difficult
, _ dust scattering is not isotropic _ : the scattering angle is smaller the shorter the wavelength of a photon , as described by the scattering phase function asymmetry parameter ( e.g. draine 2003 ; see also gordon 2004 for a recent review ) .
the magnitude of the errors introduced by simplification of the dust / stars configuration , as well as by either neglecting or approximating dust scattering , depends on the inclination of the galaxy ( baes & dejonghe 2001 ) .
in addition , _ the dusty interstellar medium ( ism ) present in the disk of a late - type galaxy is observed to be patchy _
( e.g. rix & rieke 1993 ; quillen et al .
1995 ; white , keel , & conselice 2000 ; white & keel 2001 ) .
hence , strongly reddened starlight may never dominate the uv / optical / near - ir radiation that we see .
the local structure ( or clumpiness ) of the dusty ism ( natta & panagia 1984 ; hobson & schueuer 1993 ; witt & gordon 1996 , 2000 ; bianchi , ferrara , & giovanardi 1996 ; berlind et al . 1997 ; kuchinski et al .
1998 ; bianchi et al .
2000 ) and differences in the dust / stars configuration as a function of the ages of the individual stellar populations ( silva et al . 1998
; bianchi et al .
2000 ; charlot & fall 2000 ; popescu et al .
2000 ; matthews & wood 2001 ; tuffs et al .
2004 ) affect the main result of the radiative transfer calculations , i.e. , the attenuation function . as a consequence
they affect the estimate of the total face - on extinction optical depth along the line of sight through the center of the galaxy e.g. in the v band ( 0.55 @xmath0 ) .
this quantity is referred to as the disk opacity , and offers some measure of the central dust column density and , thus , of the total amount of dust present in the disk ( e.g. misiriotis & bianchi 2002 ) .
the uncertain parameterization of the dust clumps for the milky way ( cf .
kuchinski et al .
1998 and bianchi et al .
2000 ) and other late - type galaxies leads to maximal uncertainties on the amount of dust present in a disk galaxy of the order of 40 per cent ( misiriotis & bianchi 2002 ) .
we note that the uncertainties in the dust masses of nearby galaxies determined from analyzing the far - ir / submm emission from dust ( e.g. dupac et al .
2003 ) are at least comparable to those affecting analogous estimates from extinction modeling alone ( see e.g. domingue et al .
1998 ) .
radiative transfer models for realistic dust / stars configurations simulating late - type galaxies , including multiple scattering , have been computed either via analytical approximations ( e.g. byun , freeman , & kylafis 1994 ; silva et al .
1998 ; xilouris et al .
1999 ; baes & dejonghe 2001 ; tuffs et al .
2004 ) or via monte carlo techniques ( e.g. de jong 1996 ; wood 1997 ; kuchinski et al . 1998 ; ferrara et al . 1999
; bianchi et al .
2000 ; matthews & wood 2001 ) .
in particular , kuchinski et al .
( 1998 ) made a quantitative investigation of the opacity of disks using bvrijhk photometry of 15 highly inclined sab sc galaxies and the then new dirty radiative transfer model for a bulge@xmath1disk galaxy ( gordon et al .
this set of dirty models included not only anisotropic multiple scattering but also two distinctly different , local distributions for galactic - type dust , assumed to be present in the disks of all giant late - type galaxies .
kuchinski et al .
found that the maximum optical and near - ir color excesses observed in near - edge - on dust lanes imply total central face - on extinction optical depths of 0.52.0 in the v band , consistent with previous results .
here we adopt the same family of models used by kuchinski et al . but with new parameters , as described in section 2 .
we focus on the interpretation of the behavior of the attenuation function , the color excess , and the fraction of light scattered or directly transmitted towards the outside observer for the individual bulge and disk components of a late - type galaxy like our own , as a function of the total amount of dust and the inclination of the galaxy , and the structure of the dusty ism of the disk .
in particular , the present models describe the attenuation by internal dust at far - uv wavelengths accessible through _ galex _ ( milliard et al .
preliminary results have been presented by pierini , gordon , & witt ( 2003 ) .
this study not only extends the analysis of kuchinski et al .
( 1998 ) , but also produces necessary , complementary results to those of ferrara et al .
( 1999 ) , bianchi et al .
( 2000 ) , and matthews & woods ( 2001 ) , which are based on analogous monte carlo radiative transfer simulations for disk galaxies .
in fact , ferrara et al . ( 1999 ) explored the dependence of the attenuation function on the hubble type and the inclination of a disk galaxy , as well as on the total amount of dust , the dust / stars configuration , and the extinction properties , for a large wavelength range ( 14 wavelengths from 0.125 to 2.158 @xmath0 ) , but only for a homogeneous dust distribution .
conversely , bianchi et al .
( 2000 ) investigated the effects of clumping and the presence of stars embedded in dust clumps on the total / surface photometry of disks , but only for the v band .
in addition , they assumed a different parameterization of the clumpiness of the ism and a different spatial distribution of the clumps from those assumed by kuchinski et al .
( 1998 ) or matthews & woods ( 2001 ) .
in fact , bianchi et al . ( 2000 ) derived the radial distribution of the dust clumps of their models from the co observations of the first galactic quadrant described by clemens , saunders , & scoville ( 1988 ) , showing the existence of a molecular ring , and assumed all the clumps to have the same optical depth .
conversely , clumps have a doubly exponential distribution and decreasing extinction optical depths as a function of the galactocentric distance in the models of kuchinski et al . ( 1998 ) and matthews & woods ( 2001 ) . on the other hand , matthews & woods considered , in addition to the diffuse dust distribution , a two - phase , clumpy dust distribution where the clumps have a smaller filling factor and a lower density contrast than those assumed here .
they also made computations for b , r , h , and k bands and for edge - on disks only .
finally , this study complements also the analytical description of radiative transfer presented by tuffs et al .
in fact , these authors describe the dependence of the attenuation function on the hubble type , the inclination , and the total dust mass of a late - type galaxy , for 12 wavelengths ranging from 0.091 to 2.2 @xmath0 .
they consider a dust - less stellar bulge , a disk of old stars with associated diffuse dust plus a thin disk of young stars with associated diffuse dust , and a clumpy dust component associated with star - forming regions in the thin disk ( e.g. silva et al . 1998 ) .
their description of the clumpiness approximates the star - forming regions with porous spheres , where the local absorption of photons is given by geometry ( the solid angle of a parent cloud subtended at an offspring star of given age ) , since the optical depths of the individual clouds are assumed to be extremely large at all wavelengths .
furthermore , tuffs et al .
( 2004 ) adopt values of the albedo and , especially , the scattering phase function asymmetry parameter for milky way - type dust that are sensitively different from those adopted here ( witt & gordon 200 ) .
for the calculations presented in this paper we employed the dirty radiative transfer code ( gordon et al .
the code applies to any arbitrary spatial distribution of stars and dust .
here we consider its application to the bulge and disk components of late - type galaxies separately .
with monte carlo techniques ( witt 1977 ; audic & frisch 1993 ) , photons are followed through a dust distribution and their interaction with the dust is parameterized by the optical depth , the albedo , and the scattering phase function asymmetry parameter , as defined in e.g. whittet ( 2003 ) . _
the optical depth determines the probability distribution for where a photon of given wavelength interacts ; the albedo gives the probability that the photon is scattered from a dust grain ; the scattering phase function gives the probability distribution for the angle at which the photon scatters_. for this application , the dirty code considers a dust - less stellar bulge , penetrated by a dusty , stellar disk , as customary ( see ferrara et al . 1999 and
tuffs et al .
the stellar bulge is modeled as a sphere with a stellar luminosity density decreasing exponentially with radial distance .
this is at variance with the models of both ferrara et al .
( 1999 ) and tuffs et al .
( 2004 ) , who considered bulges with @xmath2 radial profiles ( de vaucouleurs 1948 ) .
de vaucouleurs bulges are more typical of early - type spiral galaxies ( andredakis , peletier , & balcells 1995 ; graham 2001 ; hunt , pierini , & giovanardi 2004 ) . the outer radius of the bulge model is assumed to be 4 kpc , with an exponential scale length of 1 kpc , independent of the wavelength of the tracing photons .
the attenuation effects seen in the bulge are caused by a doubly exponential dust disk , which extends from the center of the bulge to the outer edge of the galactic disk component and is described later on .
the stellar disk is modeled as a doubly exponential distribution of sources with a disk radius of 12 kpc and a scale length of 3 kpc , independent of the wavelength of the tracing photons ( see tuffs et al .
2004 for a different description ) .
its total height is 2.1 kpc ( measured from the galactic plane ) but its scale height is assumed to increase with increasing wavelength of the tracing photons ( see mihalas & binney 1981 ) , as summarized in table 1 .
this variation in the stellar scale height is necessary to account for the different types of stars that dominate the flux of a galaxy at different wavelengths : the youngest stars , emitting mostly at far - uv wavelengths , are more embedded in the dusty medium than the older ones , emitting mostly in the optical / near - ir .
other models of radiative transfer deal with this aspect in different ways ( see charlot & fall 2000 ; popescu et al .
2000 ; tuffs et al . 2004 ) .
we note that the dust - to - star scale - height ratio has a larger impact on the attenuation function of a disk than the dust - to - star scale - length ratio ( ferrara et al .
1999 ) .
dust is present only in a doubly exponential disk with a full radius of 12 kpc , a scale length of 3 kpc , a scale height of 110 pc , and a total height of 2.1 kpc ( measured from the galactic plane ) , that pervades both the stellar disk and the stellar bulge .
it is assumed to consist of dust similar to that found in the average diffuse ism of the milky way ( mw ) ( witt & gordon 2000 ; see valencic , clayton , & gordon 2004 for a recent study ) , its extinction characteristics also summarized in table 1 .
it is locally distributed either in a homogeneous medium or in a two - phase clumpy one , following the algorithm of witt & gordon ( 1996 ) . in the latter case ,
_ the clump production process is stochastic with respect to the spatial coordinates of individual clumps and the smooth 3-d distribution of the bluest stars_. in particular , the volume is divided into cubic cells with sides of 44 pc , the typical diameter of a galactic giant molecular cloud ( see blitz 1991 ) , each individual cell being assigned randomly to either a high - density state or a low - density state by a monte carlo process .
the high - density - to - low - density medium contrast is set equal to 100 .
the filling factor of the clumps determines the statistical frequency of the bins in the two states ; it is set equal to 0.15 all over the dust disk . since the smooth medium has an exponential distribution , clumps close to the galactic center have a higher extinction optical depth than those in the outskirts of the dust disk . in addition , the filling factor of the clumps sets the likelihood that spatially adjoining cells are occupied by clumps , which leads to the appearence of complex cloud structures composed of one or more connected clumps .
two clumps are considered connected when they share one entire side .
the combination of the percolation approach ( see e.g. stauffer 1985 ) of generating clouds with the filling factor selected does lead to a power - law distribution of cloud sizes , as discussed by witt & gordon ( 1996 ) .
models are parameterized according to the value of @xmath3 , _ the central face - on extinction optical - depth from the surface to the mid - plane of the disk in the v band_. hence the opacity of the disk ( or the total central column density of the dust ) is equal to @xmath4 . for models with the two - phase clumpy dust distribution ,
@xmath3 is formally somewhat ill - defined ( see witt & gordon 1996 ) .
for the two - phase clumpy models , @xmath3 is the value of the face - on extinction optical depth from the surface to the center of the disk in the v band for the homogeneous model with the same total dust mass .
homogeneous and two - phase clumpy models are computed for @xmath5 , 0.5 , 1 , 2 , 4 , and 8 . the large - scale distribution of the dust within the disk coupled to the parameterization of its clumpiness on the small scale makes the disk models more optically thick in their inner regions than at their outskirts ( see kuchinski et al .
1998 ) , whatever the wavelength .
in fact , any photon traveling in a given direction through the dust disk will face at any step scattering and/or absorption according to the _ local _ value of the optical depth , which changes along both the radial and vertical coordinates as previously described .
for an evaluation of the effects on the attenuation function for a bulge@xmath1disk galaxy caused by a different dust / stars configuration or a different characterization of the clumps , we refer the reader to ferrara et al .
( 1999 ) , bianchi et al .
( 2000 ) , and tuffs et al .
( 2004 ) .
the dusty bulge@xmath1disk system is projected onto the plane of the sky , 10 different inclinations ( 0 , 18 , 36 , 50 , 65 , 70 , 75 , 80 , 85 , and 90 degrees ) being considered .
in addition , 25 wavelengths ( from 0.1 to @xmath6 ) are included , with a decimal logarithmic spacing that puts special care to sampling finely the far - uv domain accessible through _ galex _ ( milliard et al .
2001 ) . for the wide spectral range considered here , the scale height of the stellar disk ranges from 60 to 375 pc ( see table 1 ) .
one part of the solution of the radiative transfer model is the _ total attenuation optical depth _ at a given wavelength @xmath7 , which is defined as @xmath8,\ ] ] where @xmath9 is the _ total _ fraction of light at this wavelength escaping from the system , either directly or after scattering by dust , in the direction to the outside observer .
@xmath10 measures the attenuation occurring in the system , as seen by an observer from a given direction , and , thus , is _ an average over many lines of sight within the system_. since our model geometry is cylindrically symmetric , @xmath10 will depend on the inclination @xmath11 of the galaxy .
we may express the total attenuation as a function of @xmath7 in magnitudes ( a way preferred by observational astronomers ) as follows : @xmath12 = 1.086 \tau_{att,~\lambda}.\ ] ] given two pass - band filters of effective wavelengths @xmath13 and
@xmath14 , we define the color excess @xmath15 as the difference between the values of the attenuation function ( in magnitudes ) at @xmath13 and @xmath14 , respectively . here
we stress that , as a result of the radiative transfer of photons with different wavelengths through a realistic dusty medium , the attenuation function @xmath16 ( or @xmath10 ) will be different from the extinction curve assumed for the individual dust grains .
the reason is twofold .
first , radiative transfer determines both absorption and scattering , for a given extinction curve ( see e.g. gordon et al .
2003 ) , taking into account the relative distribution of stars and dust within the system and with respect to the outside observer , and the structure of the dusty medium ( see witt & gordon 1996 ) .
second , the transfer of radiation is investigated throughout the whole system , so that even lines of sight different from the observer s one may contribute to the observed radiation . for a more extensive explanation ,
we refer the reader to calzetti ( 2001 ) .
in sect . 3.1 , we investigate how the attenuation by internal dust of photons of a given wavelength , produced by stars either in the bulge or in the disk and received by the outside observer , depends on both the opacity ( or total central dust column density ) of the disk ( given by @xmath4 ) and the structure ( two - phase clumpy vs. homogeneous ) of the dusty ism of the disk . to this purpose
, we choose one specific value of the inclination @xmath11 of the bulge@xmath1disk galaxy ( i.e. , @xmath17 ) and let @xmath3 increase from 0.25 to 8 .
conversely , in sect .
3.2 , we show how much the dust attenuation for either bulge or disk , and , thus , the observed photometry of either component , depend on the view angle of the galaxy and the structure of the dusty ism of the disk , once the total amount of dust of the system and the extinction law of the mixture of dust grains are fixed .
we will discuss models corresponding to @xmath18 and @xmath11 equal to 0 , 70 or 90 degrees , since the variation of dust attenuation as a function of @xmath11 is rather slow between @xmath19 and @xmath20 but very fast for @xmath21 ( see e.g. ferrara et al .
1999 and masters , giovanelli , & haynes 2003 ) .
two different effects are expected to play a role in shaping @xmath16 , when the inclination of the galaxy increases .
one effect is the `` geometrical '' increase of the line - of - sight optical depth that a photon produced either in the bulge or in the disk faces once it is inside the dust disk on its journey towards the outside observer ( see e.g. kylafis & xilouris 1996 for the case of edge - on exponential disks ) .
the other effect is the variation of the fraction of bulge / disk light which is not affected by the dust disk owing to projection effects .
this variation is different for the bulge and disk structures .
the balance between these two competing effects depends on the wavelength , since both the extinction curve and the disk stellar distribution ( at least in our models ) depend on @xmath7 ( see sect . 2 and table 1 ) .
there is no need to assign any spectral energy distribution to either bulge or disk in order to determine the corresponding attenuation function , since for either component @xmath16 expresses a differential effect .
specification of the bulge and disk spectral energy distributions is needed when one investigates how attenuation by internal dust affects the integrated photometry of a bulge@xmath1disk system as a whole ( e.g. pierini et al .
2003 ; see also tuffs et al .
in fact , in this case one needs to count the photons produced either in the bulge or in the disk .
finally we note that the different curves reproduced in the following figures are not completely smooth and monotonic , from one wavelength included in the models to the next one , because of photon noise in the monte carlo simulations .
1 reproduces the attenuation function calculated for the bulge of a late - type galaxy ( like our own ) seen face on , as a function of @xmath3 ( half the total central dust column density of the disk ) and the structure of the dusty ism .
it also shows the extinction function expected for a homogeneous , non - scattering mw - type dust screen foreground to the light source with total v - band extinction optical depth equal to the total attenuation optical depth at v band of the face - on bulge model with a homogeneous dust disk reproduced in each panel .
this normalization is chosen in order to better show the difference in shape between the curves reproduced in fig .
1 . the comparison between these different curves reminds the non - specialized reader , if needed , that the dust obscuration for an extended object like a bulge is conceptually different from the dust extinction of stars ( see footnote 1 and e.g. calzetti 2001 ) .
2 reproduces the same set of attenuation functions displayed in fig .
1 , each one normalized to the value of the attenuation at 0.55 @xmath0 ( i.e. , @xmath22 ) . in origin , this normalization allowed the comparison of the shapes of extinction curves with different columns of dust at any wavelength . here
it is used to emphasize two aspects of the attenuation function : its slope in the far - uv spectral region and the presence of the so - called `` extinction bump '' at @xmath23 , which is a signature of the average mw - type dust ( see whittet et al .
2004 for a recent discussion ) that is used in our models .
in addition , in fig .
2 we reproduce the extinction function normalized to its v - band value expected for a homogeneous , non - scattering mw - type dust screen foreground to the light source , whatever the total v - band extinction optical depth .
this normalized extinction function is equal to the mw extinction curve normalized to its v - band value listed in table 1 .
the difference in shape between the curves reproduced in fig
. 2 is evident . from the visual inspection of fig .
1 and 2 , the following characteristics emerge . *
the average slope of the attenuation function for a face - on bulge increases together with the opacity , for @xmath24 , whatever the structure of the dusty ism .
hence , for the face - on bulge , dust reddening increases together with the disk opacity within the wavelength range where at least the inner region of the dust disk is optically thin . * the average slope of the attenuation function flattens ( steepens ) at wavelengths shorter ( larger ) than @xmath25 when the opacity increases , for @xmath26 , whatever the structure of the dusty ism .
hence , for the face - on bulge , dust reddening tends to a constant value within the wavelength range where at least the inner region of the dust disk becomes optically thick , owing to an increase in opacity .
* for a fixed @xmath7 , @xmath27 increases with increasing opacity , the rate of increase diminishing rapidly at wavelengths shorter than @xmath28 for @xmath29 , whatever the structure of the dusty ism . as a consequence , from the optical through the far - uv spectral range
, @xmath27 is almost the same for @xmath30 , so that the maximum attenuation ( at @xmath31 ) increases only from 0.5 to 1.4 mag when @xmath3 increases from 0.25 to 8 .
hence , for the face - on bulge , the increase in dust attenuation as a function of the disk opacity is strongly non - linear .
* for a fixed value of @xmath32 ( especially for @xmath33 ) , the structure of the dusty ism does not affect to zeroth order the shape of the attenuation function for the face - on bulge . *
nevertheless , a homogeneous dust distribution produces a larger attenuation than a two - phase clumpy one , whatever @xmath32 ( for @xmath34 ) .
this difference reaches a maximum factor of @xmath35 1.6 for @xmath36 . * for very low values of the opacity ( i.e. , for @xmath37 ) , the dust distribution , be it homogeneous or two - phase clumpy , has an almost negligible impact on the attenuation function for the face - on bulge . slight discrepancies ( of the order of 0.15 magnitudes or less ) are present only shortward of @xmath38 ( i.e. , rightward of @xmath39 ) . * at wavelengths shorter than @xmath28 , the discrepancy between the homogeneous case and the two - phase clumpy case increases with increasing opacity , for @xmath29 , but decreases with increasing opacity , for @xmath40 . *
conversely , at wavelengths longer than @xmath25 ( i.e. , leftward of @xmath41 ) , the discrepancy between the homogeneous case and the two - phase clumpy case increases with increasing opacity along the whole range of @xmath3 explored by the models . * the relative prominence of the @xmath42-extinction
bump diminishes when @xmath3 increases .
thus , radiative transfer effect may weaken this feature in far - uv spectra of face - on bulges of nearby galaxies .
this behavior may be explained as follows .
first we recall that , by construction ( see sect . 2 ) , the face - on projection of the bulge is almost coincident with the face - on projection of the dustiest region of the disk ( within one scale length ) , where the _ local _ optical depths are thus the largest .
for this projection , almost half of the bulge stands between the dust disk and the outside observer .
the light produced in the face - on bulge is effectively removed from the observer s line of sight when the amount of dust increases , especially when it is emitted at @xmath43
. however , the increase of @xmath27 together with @xmath3 slows down when a _
local _ optical depth of order 1 is approached across the disk region corresponding to the face - on projection of the bulge ( first at the shortest wavelengths and then at progressively larger ones ) . in the ideal case of a bulge bisected by an opaque dust layer
, we expect that this dust blocks _ exactly _ half of the light emitted by the bulge towards the outside observer ( i.e. , @xmath44 for any @xmath7 ) , when we look at it from a direction perpendicular to the dust layer .
obviously , this is not the dust distribution of our models , so that the shape of the attenuation function for the face - on bulge changes as a function of @xmath3 , while @xmath27 approaches a sort of asymptotic value , when @xmath3 increases from 0.25 to 8 .
this causes the simultaneous flattening of the attenuation function , from the shortest wavelengths towards progressively longer ones , for @xmath26 , whatever the structure of the dusty ism of the disk .
the asymptotic value is larger than 0.75 since the dust / stars configuration of our model does not resemble the ideal case previously described . the simulations for the face - on bulge confirm that both absorption and scattering diminish in a two - phase clumpy medium , in comparison with a homogeneous medium having the same total mass and type of dust ( see witt & gordon 1996 for a detailed analysis ) .
it is for this reason that the attenuation of the light produced in the face - on bulge at a given wavelength becomes `` grey '' ( i.e. , tends to an asymptotic value ) at slightly lower values of @xmath3 for a homogeneous dust distribution than for a two - phase clumpy one .
_ however , the structure of the dusty ism of the disk produces only negligible effects on the observed colors of a face - on bulge .
_ as a further result of the radiative transfer calculations , we find that the absorption excess represented by the @xmath42-extinction bump ( calzetti et al .
1995 ) is washed out when @xmath3 increases .
this effect was noted first by cimatti et al .
( 1997 ) for a different dust / stars configuration from that assumed here .
what happens in fig .
1 and 2 is that , for photons of the bulge emitted at neighbouring wavelengths with respect to @xmath42 , chances for both ( multiple ) scattering and the eventual absorption increase together with @xmath3 , so that the attenuation reaches an asymptotic value at and around @xmath42 .
fig . 3 and 4
reproduce the attenuation function and the normalized attenuation function , respectively , for the disk of a late - type galaxy seen face on , as a function of the opacity and the structure of the dusty ism of the disk , in analogy with fig . 1 and 2 , respectively , for the bulge . from the visual inspection of fig .
3 and 4 , we draw the following conclusions . * the average slope of the attenuation function for a face - on disk increases with increasing opacity , through the whole range of @xmath3 considered here , whatever the structure of the dusty ism . hence
, for the face - on disk , dust reddening is a monotonically increasing function of the opacity .
* however , the attenuation optical depth increases at a reduced rate , at least in the uv spectral region , when @xmath3 becomes larger than 1 , so that the maximum attenuation ( at @xmath31 ) increases from 0.2 to 2.6 mag ( for the homogeneous dust distribution ) when @xmath3 increases from 0.25 to 8 . *
a difference in the shape of the attenuation function persists between the homogeneous case and the two - phase clumpy case , whatever @xmath3 ( for @xmath34 ) . * also for a face - on disk the homogeneous dust distribution produces a larger attenuation than a two - phase clumpy one , at any wavelength .
this time the discrepancy in the values of @xmath27 produced by the two very different structures of the dusty ism increases with increasing total central dust column density , whatever the wavelength . * the absorption - excess feature at @xmath23 can be easily identified , though its prominence diminishes with increasing opacity ( for @xmath34 ) . in order to understand this behavior ,
it is necessary to refresh first some properties of the dust / stars configuration of our disk models .
the distribution of dust and stars along the radial coordinate of the disk is assumed to be described by the same exponential law , whatever the parent stellar population of the stars .
conversely , the vertical distribution of the different stars of the disk depends on their parent stellar population .
thus the scale height decreases together with the wavelength characterizing to zeroth order the age of the individual stars ( see sect . 2 and table 1 ) .
as a consequence , the fraction of light which is produced in the disk at a given wavelength and is affected by the dust , depends even more on @xmath7 , with respect to the case of a vertical distribution with a unique scale height ( as e.g. in ferrara et al .
1999 ) , for fixed extinction properties , total amount of dust , and structure of the dusty ism of the disk .
having assumed the dust / stars configuration for the disk as in sect .
2 , we note that the light produced by the stars in the outskirts of the disk will shine almost unattenuated for very low values of the disk opacity , whatever the wavelength
. however , when @xmath3 increases , a _ local _ optical depth @xmath45 or larger will be eventually reached within a progressively thicker region , delimited by a progressively larger radius ; the longer @xmath7 the smaller this region .
it is now straightforward to understand why , for the face - on disk , @xmath27 increases together with @xmath3 , at any given @xmath7 , and why this increase is so dramatic at @xmath46 ( see table 1 ) .
the best example of the importance of the dust / stars configuration for the result of the radiative transfer analyzed here is offered by the far - uv photons emitted at @xmath47 , since they are produced by the stars the most deeply embedded in the dust layer ( see table 1 ) .
these photons might eventually never reach the outside observer if @xmath3 were allowed to increase indefinitely , whatever the height from the galactic plane and radius where they are produced .
conversely , optical / near - ir photons may still reach the observer in large number even for @xmath48 .
for photons emitted at a fixed wavelength , a constant value of the attenuation equal to 0.75 mag is expected when we see half the light from the face - on disk bisected by an opaque dust layer .
since the dust / stars configuration matters , our face - on disk models produce an attenuation of less than 0.75 mag for photons emitted shortward of @xmath49 even for @xmath50 .
we also note that the attenuation at uv wavelengths almost doubles its initial value when @xmath3 increase from 0.25 to 0.5 , and from 0.5 to 1 . at those wavelengths ,
the increase of @xmath27 together with @xmath3 slows down for larger values of @xmath3 . despite that
, the absorption - excess feature at @xmath23 may still be easily identified in the attenuation function for the face - on disk model with @xmath48 , though its prominence has become modest .
information on this feature is carried by the light escaping from the regions of the disk where the _ local _ optical depth is low enough ( e.g. in the outskirts of the disk ) .
finally , we note that the attenuation in the k band assumes a slightly negative value in the case of a face - on disk with @xmath37 ( or less ) .
this means that the outside observer receives more k - band photons from the face - on disk than it would be in absence of dust .
the reason for this excess is the contribution of those k - band photons initially traveling in different directions but scattered by the dust towards the outside observer .
this effect has already been highlighted by baes & dejonghe ( 2001 ) . from the visual inspection of fig . 1 through 4 it is clear that the solutions of the radiative transfer for bulge and disk of a late - type galaxy seen face on are different .
the reason is the different distribution of stars and dust for either system with respect to the outside observer .
the face - on projection of the bulge is always within the dustiest region of the disk ( by construction ) , and , if the opacity is large enough , the fraction of stars whose light is significantly affected by the dust is the same whatever the emission wavelength .
conversely , for the same value of the opacity , the face - on projection of the disk will present regions almost unaffected by the dust at larger galactocentric distances the shorter the wavelength of the emitted photons . at the same time , the fraction of stars of the disk whose light is significantly affected by the dust is larger the shorter the emission wavelength , whatever the radial distance ( also by construction ) .
of course , for both bulge and disk , geometry is convolved with the selective nature of absorption and scattering in the computation of the attenuation function . here
we discuss the behavior of the optical color excess @xmath51 and the near - ir color excess @xmath52 , as defined in sect .
2 , as a function of the opacity and the structure of the dusty ism of the disk . this exercise offers a very useful example of how confusing the extinction law of the mixture of dust grains present in the dusty ism of a galaxy with the attenuation function of the galaxy itself may lead to extremely wrong conclusions regarding the color excess expected for this dusty stellar system .
furthermore , we discuss whether the ratio of total to selective _ attenuation _
@xmath53 for a dusty stellar system , like a late - type galaxy , represents a diagnostic of the total - to - selective _ extinction _ ratio of the mixture of dust grains that this system contains .
the values of the attenuation function at the effective wavelengths of the filters b , v , j , and k ( i.e. , at @xmath54 0.44 , 0.55 , 1.25 , and @xmath55 , respectively ) are determined from the linear interpolation of the models displayed in fig . 1 and 3 . for the face - on bulge , @xmath51 reaches an absolute maximum of about 0.1 mag for @xmath36 and decreases down to @xmath56 mag for larger values of the opacity in the homogeneous case , while it never exceeds @xmath57 mag ( for @xmath34 ) in the two - phase clumpy case ( see fig .
conversely , @xmath52 increases non - linearly with @xmath3 up to @xmath58 mag or @xmath59 mag , for the homogeneous or two - phase clumpy dust distribution , respectively , the increase being slower in the latter case ( see fig .
5b ) . note that , for both local distributions of the dust , @xmath52 is definitely lower than @xmath51 only for @xmath24 .
this behavior is surprising only at first glance .
in fact we can understand it when considering the results reproduced in fig . 1 and 2 .
there it is shown that @xmath60 is slightly greater than @xmath22 and increases faster than @xmath22 for @xmath61 , especially in the homogeneous case .
however , @xmath62 for larger values of the opacity , when the number of lines of sight with optical depths at b- and v - band less than unity is very much reduced across the region of the dust disk crossed by photons originated in the bulge .
this regime is reached at a slightly lower value of @xmath3 ( where the turn - around of @xmath51 takes place ) for the homogeneous dust distribution than for the two - phase clumpy one .
@xmath51 never reaches values larger than 0.1 mag when @xmath3 increases beyond @xmath63 since the attenuation function at optical wavelengths is rather flat for @xmath64 .
thus , we conclude that the amount of reddening of a ( face - on ) bulge , as measured by @xmath51 , is not a measure of the opacity of the associated dust disk , except for low values of @xmath3 .
conversely , fig . 1 and 2
show that @xmath65 is greater than @xmath66 , and increases faster than @xmath66 for @xmath67
. however , the attenuation function tends to flatten even in the near - ir for @xmath68 . as a consequence ,
@xmath52 is a monotonically increasing function of @xmath3 , for the whole range of opacities sampled here , but it is expected to show a behavior similar to that of @xmath51 if even larger values of the opacity were considered . furthermore , fig . 1 and 2
show that the attenuation function of a face - on bulge is steeper in the near - ir than in the optical , for @xmath64 .
for this reason , @xmath52 may reach larger values than @xmath51 .
the decrease of @xmath69 towards larger values of @xmath3 ( see fig .
6a ) is the consequence of the previously described behavior of the attenuation optical depth at each of the four effective wavelengths as a function of the amount of dust . fig .
6b shows another result which is surprising only at first glance : for the face - on bulge , @xmath70 spans a wide range of values ( from @xmath71 to @xmath72 and from @xmath71 to @xmath73 , in the homogeneous and two - phase clumpy cases , respectively ) as a function of @xmath3 . as a result of the radiative transfer , @xmath70 , here defined as the ratio of the _ attenuation _ at @xmath74 to the color excess @xmath51 , significantly departs from the value of 3.1 , which is the total - to - selective _ extinction _ ratio for the average mw - type extinction law ( see e.g. valencic et al .
the non - linear dependence on the total amount of dust of @xmath75 reflects the flattening of the attenuation function with increasing @xmath3 ( fig .
2 ) . for the face - on disk , @xmath51 and @xmath52
are monotonically increasing functions of the opacity , at least for @xmath34 ( see fig .
however , we note that a variation of @xmath51 as small as 0.02 mag may correspond to a variation of @xmath3 by a factor of 2 . thus we conclude that also the amount of reddening of a face - on disk , as measured by @xmath51 , is not a sensitive measure of the opacity of the disk .
7 also shows that the optical and near - ir color excesses are larger in the homogeneous case than in the two - phase clumpy one .
furthermore , @xmath51 is greater ( lower ) than @xmath52 for @xmath76 ( @xmath77 ) . as a consequence
, @xmath69 drops from @xmath78 to the asymptotic value of @xmath79 ( see fig .
8a ) , the dynamic range of this drop being similar to that observed for the face - on bulge ( from @xmath80 to @xmath81 ) in fig .
6a . once again , the behavior of @xmath69 for the face - on disk reflects the change of the shape of the attenuation curve as a function of @xmath3 ( fig .
4 ) . finally , for the face - on disk , @xmath70 increases from @xmath82 up to the asymptotic value of @xmath83 ( @xmath84 ) in the homogeneous ( two - phase clumpy ) case , when @xmath3 increases from 0.25 to 8 ( fig .
this is a consequence of the steepening of the attenuation function in the optical spectral region with increasing opacity . from fig .
6b and 8b we conclude that radiative transfer effects hinder the evaluation of the total - to - selective _ extinction _ ratio , typical of the dust mixture present in a dusty system like a bulge@xmath1disk galaxy , from the total - to - selective _ attenuation _ ratio observed for either bulge or disk . here
we discuss in detail the fraction of light , emitted at a given wavelength ( from the far - uv through the near - ir ) in the bulge or disk component of a face - on late - type galaxy , which is either scattered by dust or directly transmitted towards the outside observer . for the face - on bulge , the fraction of light scattered towards the outside observer ( @xmath85 ) never exceeds 20 per cent of the light initially emitted along the observer s line of sight , for the range of opacity investigated here ( see fig .
the wavelength where this fraction peaks increases when @xmath3 increases .
shortward of this peak wavelength , the fraction of scattered light decreases with increasing disk opacity . as a consequence ,
the color of the scattered light is blue for @xmath86 and red for larger values of @xmath3 .
this behavior may be easily understood when considering that the efficiencies of both absorption and ( multiple ) scattering at a given wavelength increase when @xmath3 increases .
when @xmath3 increases , ( multiple ) scattered far - uv photons originated in the face - on bulge will be eventually absorbed by the face - on dust disk ; conversely , optical / near - ir photons originated in the face - on bulge will be scattered more efficiently in the direction to the outside observer , initially , but absorbed in increasing proportion as soon as the _ local _ optical depths at these wavelengths equal unity or become larger across the region of the dust disk crossed by those photons . conversely , the fraction of light directly transmitted towards the outside observer ( @xmath87 ) is a monotonically decreasing function of @xmath3 , whatever the wavelength ( see fig .
this fraction becomes constant shortward of a break wavelength for @xmath26 , the break wavelength increasing with increasing values of the opacity .
longward of this wavelength the fraction of light directly transmitted along the observer s line of sight is larger by up to a factor of two . as a consequence of the previous behavior , the scattered - to - directly transmitted light ratio ( @xmath88 ) increases ( decreases ) with decreasing wavelength longward ( shortward ) of the break wavelength
( see fig .
this ratio never exceeds 35 per cent at any given @xmath7 . finally , we find that the clumpiness of the dusty ism reduces both absorption and scattering of the light emitted by the face - on bulge , whatever the amount of dust .
however , for a face - on bulge , @xmath88 is only mildly ( if not negligibly ) dependent on the clumpiness of the dusty ism of the disk . for the face - on disk
, @xmath85 increases together with @xmath3 , for @xmath29 , while it decreases when @xmath3 becomes larger , whatever the wavelength ( see fig .
the peak of the scattered light is in the far - uv spectral region for values of @xmath3 as large as 4 ; it moves to the optical / near - ir spectral region for larger values of @xmath3 .
the interpretation of this behavior is the same as for the face - on bulge .
conversely , @xmath87 is a monotonically decreasing function of @xmath3 , whatever the wavelength , and no break wavelength is present ( see fig .
this may be easily understood from the discussion in sect .
in addition , @xmath88 increases together with the opacity , whatever the wavelength .
this is due to the dramatic decrease of @xmath87 when @xmath3 increases , especially for the far - uv wavelengths .
the increase of @xmath88 as a function of @xmath3 strongly depends on the clumpiness of the dusty ism ( see fig .
14 ) . furthermore ,
@xmath88 peaks at about 0.15 @xmath0 whatever @xmath3 ( for @xmath34 ) .
we note that , for @xmath48 , @xmath85 is about 50 ( 30 ) per cent of the 0.15 @xmath0-flux reaching the observer in the homogeneous ( two - phase clumpy ) case .
there is a lot of difference concerning the behavior of @xmath88 between the face - on disk and the face - on bulge .
once again the reason is the different configuration of dust and stars emitting at a given wavelength for the bulge and the disk .
15 shows the complex behavior of @xmath27 for the bulge of a galaxy with a dust disk having @xmath18 and mw - type dust , as a function of the inclination of the galaxy .
the attenuation optical depth increases faster at optical wavelengths than at far - uv ones , when @xmath11 increases from @xmath19 to @xmath20 , as can be realized from fig .
16 , where @xmath90 is displayed .
the final result is that the attenuation function flattens shortward of @xmath91 when @xmath92 . at the same wavelengths ,
the increase is reversed and the flattening accentuated for @xmath11 @xmath93 .
conversely , longward of @xmath91 , the attenuation optical depth increases proportionally to @xmath94 , with the result that the attenuation function steepens , when @xmath11 increases from @xmath19 to @xmath20 . for higher inclinations ,
the attenuation optical depth increases faster at the largest near - ir wavelengths than at the shortest ones , so that @xmath27 tends to flatten also in the near - ir .
when the inclination of the galaxy becomes larger , a photon of given wavelength , emitted in the bulge and traveling through the dust disk on its way to the outside observer , faces a larger number of lines of sight with optical depths greater than unity whatever @xmath7 . on the other hand ,
a larger fraction of the bulge comes into the view of the outside observer unobstructed by the dust disk , when @xmath11 increases .
in fact , the bulge model has a ( wavelength - independent ) scale height larger than that of the dust distribution by almost a factor of 10 , and a radial extension 3 times shorter than that of the dust disk ( cf .
2 ) . fig .
15 shows that the increase of the projected surface of the bulge which is unobstructed by the dust disk prevails over the competing increase of the _ local _ optical depth through the dust disk , for all wavelengths except the longest near - ir ones . at the same time , the attenuation function for the edge - on bulge is grey throughout almost the whole spectral region under consideration .
for this reason , the absorption peak at @xmath23 , typical of the mw - type extinction law , is almost undetectable in the attenuation function of the edge - on bulge . from the previous discussion ,
it is clear that the blocking action of the dust disk must peak at some intermediate inclination , since almost half of the bulge stands between the dust disk and the observer , in the face - on case , and most of the projected surface of the bulge is unobstructed by the dust disk , in the edge - on case . the value of the inclination which maximizes the blocking action of the dust disk will depend on the wavelength of the emitted photons .
in addition to the previous two geometric effects , the wavelength - dependence of the scattering phase function asymmetry parameter ( see table 1 ) plays a role in interpreting fig .
15 and 16 .
we recall that scattering is more forward directed the shorter the wavelength of the scattered photon .
thus the near - ir photons emitted in the bulge and traveling through the dust disk will have a greater probability of being scattered out of the observer s line of sight than the optical / far - uv ones , in the edge - on case .
this contributes to explaining why the attenuation by internal dust of the light produced in the bulge at near - ir wavelengths increases monotonically together with the inclination of the galaxy .
finally , we note that the clumpiness of the dusty ism of the disk ( with @xmath95 ) produces distinguishable effects on the attenuation function for the bulge only for the lowest inclinations of the galaxy , i.e. , when @xmath27 is the lowest whatever @xmath7 .
this is explained in the following section .
the dependence on inclination of the attenuation function for a disk with @xmath18 and mw - type dust is very dramatic , as shown in fig .
in fact , the attenuation increases ( almost doubling its initial value ) at any @xmath7 , when @xmath11 increases from @xmath19 up to @xmath20 , owing to the increasing probability of absorption for a photon of any wavelength traveling through the dust disk towards the outside observer . for higher inclinations
, the attenuation function becomes suddenly very steep , with the result that @xmath96 in the edge - on case . for inclinations close to @xmath97 ,
the photons produced in the dustiest region of the disk and traveling towards the outside observer will be eventually absorbed there , whatever their wavelength .
the photons produced in the peripheral ( and less dusty ) regions of the disk and traveling towards the outside observer will experience significantly increased values of the line - of - sight optical depth with respect to the face - on case .
they may be scattered out of the observer s line of sight , if their wavelength belongs to the optical / near - ir domain .
conversely , photons produced at higher heights above the disk will have a greater probability to reach the outside observer . since the ratio of stellar disk scale - height and dust disk scale - height is an increasing function of the emission wavelength , by construction ( see sect . 2 and table 1 )
, absorption will be even more dramatic the shorter the wavelength .
in addition to the previous geometrical effect , the wavelength - dependence of the scattering phase function asymmetry parameter ( see table 1 ) predicts that ( multiple ) scattered photons will be eventually absorbed in larger fraction the shorter their wavelength , as already discussed .
this explains the abrupt rise of @xmath27 at far - uv wavelengths for inclinations close to @xmath97 . as a consequence , for these inclinations , the @xmath42-bump , typical of mw - type dust , is almost undetectable ( see fig .
17 and 18 ) .
finally , we note that the impact of the structure of the dusty ism on the attenuation function , in terms of absolute values and shape , is almost negligible when the disk is seen at @xmath98 . for such high inclinations
, the optical depth along the observer s line of sight may be much larger than unity for many if not most of the wavelengths just because many dust clumps line up in that direction .
hence the eventual absorption of a photon of given wavelength happens just by virtue of the total cross section ( or blocking effect ) of the dust clumps .
this result complements that of matthews & woods ( 2001 ) , who noted that differences in the structure of the dusty ism , for their edge - on models , impact sensitively on the values of the attenuation but not on the shape of the attenuation function . for the bulge models associated with a dust disk having @xmath18 , fig .
19a shows that either @xmath51 is constant ( homogeneous dust distribution ) or it slowly increases together with the inclination ( two - phase clumpy medium ) , for @xmath99 @xmath11 @xmath100 .
however , @xmath51 declines for higher inclinations , whatever the local distribution of the dust . as a consequence , for the bulge @xmath51 is bluer ( by 0.04 mag ) in the edge - on case than in the face - on case .
conversely , @xmath52 is a strongly non - linear function of the galaxy inclination : it increases from @xmath101 to @xmath59 mag , when @xmath11 increases from @xmath19 to @xmath102 ; it decreases for higher inclinations , whatever the dust distribution ( see fig .
the initial increase of @xmath52 is a consequence of the steepening of the attenuation function for the bulge in the near - ir , when @xmath11 increases from @xmath19 to @xmath20 , which means that the dusty medium crossed by the photons emitted in the bulge and traveling towards the outside observer tends to become optically thick also in this wavelength domain .
the decrease of both @xmath51 and @xmath52 for inclinations approaching @xmath97 is due to the fact that an increasing fraction of the light produced in the bulge reaches the outside observer without being affected by the dust ( within the disk ) .
the turn - over in the behavior of the color excess happens when the increase in the line - of - sight optical depth faced by the photons traveling through the dusty medium is compensated by the reduction in the number of such photons .
it occurs at higher inclinations for @xmath52 than for @xmath51 because of the wavelength - dependence of the extinction law ( see table 1 ) . as a consequence of the previous behavior
, @xmath69 decreases monotonically from 1.7 to 0.3 when @xmath11 increases from @xmath19 to @xmath97 ( see fig .
thus , for the bulge , @xmath52 is almost three times larger than @xmath51 for @xmath11 @xmath102 .
conversely , @xmath70 is almost constant ( @xmath103 ) , for @xmath104 , but increases for higher inclinations ( up to @xmath105 for the edge - on bulge ) without any significant dependence on the local distribution of the dust ( see fig .
20b ) . for a disk with @xmath18
, both @xmath51 and @xmath52 are monotonically increasing functions of @xmath11 , @xmath51 being larger than @xmath52 for @xmath11 @xmath106 ( see fig .
this is a consequence of the fact that @xmath16 increases monotonically together with @xmath11 ( cf .
17 ) , the increase being slower in the optical than in the near - ir when inclinations higher than @xmath20 are approached .
in addition , we find that the clumpiness of the dusty ism does not affect the behavior of the near - ir color excess , but produces a slight difference for the optical one , as expected from fig . 17 . as a consequence of the previous behavior
, @xmath69 is a monotonically decreasing function of @xmath11 , ranging from @xmath71 ( face - on disk ) to @xmath45 ( edge - on disk ) , as shown in fig .
this range is similar to that obtained for the bulge ( cf .
conversely , @xmath70 is a slowly increasing function of @xmath11 ( see fig .
22b ) , ranging from @xmath71 ( face - on disk ) to @xmath84 ( edge - on disk ) .
this confirms that , for the disk , the attenuation at optical wavelengths increases slightly faster than the reddening at the same wavelengths , even for @xmath107 ( cf .
17 and 18 ) .
this is due to the geometrical increase of the region of the disk which is opaque to optical photons .
the fraction of light which is produced in the bulge and is scattered by the dust disk towards the outside observer diminishes when the inclination of the galaxy increases ( see fig .
it ranges from @xmath35 20 per cent ( face - on case ) to 5 per cent ( edge - on case ) . for the case of @xmath18 considered here , the color of the scattered light is always red and the clumpiness of the dusty ism does not play an important role .
conversely , the fraction of light which is produced in the bulge and is unaffacted by the dust ( and , thus , reaches directly the outside observer ) has a non - monotonic behavior as a function of the inclination of the galaxy ( see fig .
24 ) . for @xmath108 ,
this fraction decreases with increasing @xmath11 , whatever @xmath7 . for higher inclinations , a minor fraction of the dust disk projects into the bulge , so that @xmath87 increases again in the optical / far - uv spectral domain , where absorption already prevails over scattering for @xmath95 .
the reason why @xmath87 keeps on decreasing in the near - ir , for @xmath98 , although a larger fraction of the bulge comes into the view of the outside observer without being obstructed by the dust disk , is that both absorption and scattering out of the observer s line of sight become more effective in the near - ir , for @xmath18 .
the effect on @xmath87 of the clumpiness of the dusty ism diminishes for increasing inclinations , owing to the blocking effect of the dust clumps along the observer s line of sight , so that the two model curves in fig .
24 are undistinguishable for @xmath11 @xmath109 . as a result of the behaviors previously described , for the bulge @xmath88 dimishes from 35 per cent ( at most ) to 15 per cent ( at most ) when the inclination of the galaxy increases from @xmath110 to @xmath97 ( see fig .
when the inclination of the galaxy increases from @xmath110 to @xmath97 , the photons produced in the disk and traveling through its dusty ism will experience larger line - of - sight optical depths on their way towards the outside observer , simply because the average path length within the dust distribution increases .
this implies that the probability of multiple scattering events rises for increasing inclinations , as well as the probability that the multiple - scattered photons are eventually absorbed inside the dust disk .
thus , @xmath85 and @xmath87 are expected to monotonically decrease for any @xmath7 when @xmath11 increases , as shown in fig .
26 and 27 , respectively .
we note that the color of the scattered light turns from blue to red only for the highest inclinations , when the far - uv photons are absorbed for the most part .
nevertheless , in the edge - on case , the scattered photons make up to 30 per cent of the total far - uv photons which reach the outside observer ( see fig .
28 ) . in the optical and near - ir
, @xmath88 slightly increases for @xmath111 @xmath11 @xmath112 , but drops for higher inclinations , when absorption prevails over scattering along the observer s line of sight , and the scattered photons which manage to leave the system do so mostly in other directions or from the peripheral regions of the disk , where the _ local _ optical depth is low .
the interpretation of the spectral energy distribution observed for a dusty stellar system requires understanding the radiative transfer problem for that particular system . on the other hand , the solution given by a radiative transfer model ( e.g. the attenuation function , investigated here , which describes the behavior of the total attenuation optical depth as a function of wavelength ) depends on the dust / stars configuration , the total amount of dust , the extinction properties of the mixture of dust grains ( e.g. gordon et al .
2003 ) , and the structure of the dusty interstellar medium ( ism ) assumed for the system .
these aspects have been extensively discussed by e.g. witt & gordon ( 1996 , 2000 ) for the spherically symmetric case .
different descriptions of the dust / stars configuration and/or the structure of the dusty ism are adopted by the existing different calculations of the attenuation function ( including multiple scattering ) for a bulge@xmath1disk galaxy , whether these computations are performed analytically ( e.g. byun et al .
1994 ; silva et al . 1998 ; xilouris et al . 1999 ; baes & dejonghe 2001 ; tuffs et al . 2004 ) or via monte carlo techniques ( e.g. de jong 1996 ; wood 1997 ; kuchinski et al . 1998 ; ferrara et al . 1999
; bianchi et al .
2000 ; matthews & wood 2001 ) .
some of these authors discuss in detail the further dependences of the attenuation function on the inclination , the structure of the dusty ism ( bianchi et al .
2000 ; matthews & wood 2001 ) , and the bulge - to - disk ratio of the galaxy ( ferrara et al . 1999 ; tuffs et al . 2004 ) .
we present new monte carlo simulations for the attenuation by internal dust of the light produced either in the bulge or in the disk of a giant late - type galaxy like the milky way , based on the dirty code of radiative transfer of stellar and scattered radiation ( gordon et al .
these models do not include the absorption and scattering of photons associated with all recombination lines _ in h ii regions themselves _ , as do those of silva et al .
( 1998 ) , panuzzo et al .
( 2003 ) , and tuffs et al .
. however they do include the absorption and scattering of photons associated with all recombination lines _ after they emerge from the h ii regions_. in addition , they explore a different and important region of the parameter space , concerning both the description of the clumpiness of the dusty ism of the disk , and the assumed 3-d spatial distribution of the sources of photons of fixed wavelength @xmath7 with respect to the 3-d spatial distribution of the dust .
these aspects are discussed in sect .
we recall that our models assume a purely stochastic distribution of the dust clumps , with respect to each other and the bluest stars .
the dust clumps are simulated as cubic cells with a size of 44 pc , and a filling factor set equal to 0.15 all over the dust disk , while the clumpy - to - diffuse ism phase density ratio is set equal to 100 .
this set of parameters was found by witt & gordon ( 1996 ) to be closest to that representing the structured ism of the galaxy . for a spherically symmetric geometry
, it represents an intermediate case between the case where the interclump medium controls the optical depth of the system , with results very similar to the homogeneous case , and the case where the optical depth of the system is controlled by the blocking effect of the clumps and is then proportional to the filling factor ( see witt & gordon 1996 ) .
since the diffuse ism of the disk models used here has a doubly exponential distribution , clumps close to the center of the galaxy have higher opical depths than those at the periphery of the disk .
conversely , the dust clumps in the monte carlo simulations of bianchi et al .
( 2000 ) are distributed mostly in a ring structure ( clemens et al .
1988 ; see also sodroski et al .
1997 ) and have all the same optical depth . in addition , the description of the clumpiness of the dusty ism of the disk given by bianchi et al .
( 2000 ) corresponds to quite low global filling factors , for the three values of the fraction of the total gas mass attributed by the latter authors to the molecular component of the ism .
bianchi et al . and misiriotis & bianchi ( 2002 ) have already discussed the differences in the behavior of the attenuation for the disk , which arise as a consequence of these two different parameterizations of the clumpy ism , as a function of the inclination of the galaxy .
a low filling factor of the clumps ( equal to 0.06 ) and a density contrast of 20 seem to reproduce better the structured ism of low surface brightness galaxies like ugc 7321 ( matthews & wood 2001 ) .
the different set of parameters describing the clumpiness of the dusty ism in the milky way ( witt & gordon 1996 ) and in the low rotational velocity sd galaxy ugc 7321 may reflect systematic changes in the turbulent velocities supporting the gas layer of a disk galaxy as a function of rotational velocity ( dalcanton , yoachim , & bernstein 2004 ) .
finally , constant , large optical depths ( whatever the wavelength ) are assumed by tuffs et al .
( 2004 ) to characterize the clumpy distribution of dust in their models , that is associated with the opaque parent molecular clouds of massive stars ( see also silva et al . 1998 ) .
the attenuation produced by this clumpy component of the ism does not depend on the inclination of the galaxy ( tuffs et al .
its wavelength dependence is due to the different blocking action of the clumps on the photons emitted by stars of different masses ( and , thus , ages ) .
stars of different masses survive for different times , so that lower mass , redder stars can escape further away from the star - forming complexes during their longer lifetimes .
the models under investigation here include only milky way - type dust ( witt & gordon 2000 ) , while ferrara et al .
( 1999 ) consider also small magellanic cloud - type dust ( witt & gordon 2000 ) .
the values of albedo and , especially , asymmetry of the scattering phase function for milky way - type dust given by witt & gordon and listed in table 1 differ sensitively from those adopted by tuffs et al .
( 2004 ) .
the doubly exponential disk models considered here assume that the stars emitting their bulk luminosity at a given wavelength have a larger scale height the longer the wavelength of the tracing photons ( see table 1 ) , but the same scale length whatever the wavelength .
this is at variance with the doubly exponential disk models of ferrara et al .
( 1999 ) and matthews & wood ( 2001 ) , where the dust - to - stars scale - height ratio is independent of the wavelength .
conversely , tuffs et al .
( 2004 ) assume that the old and young stellar populations of the disk are distributed with two different scale heights but with the same scale length , the latter increasing the shorter the wavelength .
in addition , in their disk models old and young stars have different dust - to - stars scale - height ratios , independent of @xmath7 .
these differences reflect the present ambiguous knowledge of the global structure of disks .
structural studies of edge - on disk galaxies seem to reveal the ubiquity of thick disks with red , old stellar populations similar to the thick disk of the milky way ( e.g. de grijs & van der kruit 1996 ; dalcanton & bernstein 2002 ) , but the magnitude of the intrinsic vertical color gradients is uncertain ( e.g. de grijs & peletier 2000 ) . in addition , it is not established yet if the wavelength dependence of the scale lengths of a disk galaxy represents also the existence of stellar population gradients along the radial direction ( e.g. de jong 1996 ; kuchinski et al . 1998 ; xilouris et al . 1999 ) or is only due to dust attenuation ( e.g. de grijs 1998 ) . finally ,
the bulge models under investigation here have exponential radial profiles against the de vaucouleurs one included by ferrara et al .
( 1999 ) and tuffs et al .
( 2004 ) . in all three cases
the typical scale of the bulge does not depend on the wavelength however . in the previous sections ,
we have illustrated how the attenuation function , the color excess , and the fraction of light scattered towards the outside observer behave as a function of the total central face - on extinction optical - depth at v band ( i.e. , the opacity , equal to @xmath4 in our models ) of the disk , the inclination of the galaxy , and the structure ( two - phase clumpy vs. homogeneous ) of the dusty ism of the disk .
as a first result , we confirm that _ the attenuation function for either bulge or disk changes dramatically as a function of both the disk opacity _ ( see sect .
3.1.1 ) , regulating the total amount of dust of the models , _ and the inclination @xmath11 of the galaxy , for fixed dust / stars configuration and structure of the dusty ism of the disk _ ( see sect .
this has been already described by ferrara et al .
( 1999 ) , pierini et al .
( 2003 ) , and tuffs et al .
it is no surprise : it is a direct consequence of the definition of the attenuation function as the combination of the extinction curve with the geometry of a dusty stellar system .
the extinction curve describes the combined absorption and out - of - the - beam scattering properties of a mixture of dust grains of given size distribution and chemical composition in a screen geometry as a function of wavelength .
it is an intrinsic property of the system , together with the total amount of dust , the 3-d distribution of dust and stars , and the structure of the dusty ism .
conversely , geometry depends also on the view angle of the system , since the aspect of the dust / stars configuration changes as a function of the observer s line of sight . for all these reasons
the behavior of the attenuation as a function of the opacity and the inclination of a bulge@xmath1disk galaxy is expected to be different for bulge and disk . as a consequence ,
we find , not surprisingly ( see e.g. calzetti 2001 ) , that the attenuation function for either bulge or disk is very different from the extinction function expected for a homogeneous , non - scattering mw - type dust screen foreground to the light source .
the same applies to any color excess . for the bulge of a face - on late - type galaxy , the attenuation at a fixed wavelength @xmath7 increases with increasing @xmath3 , while the attenuation function becomes increasingly flat ( or `` gray '' ) at those wavelengths where the line - of - sight optical depth is larger than unity across most of the region of the dust disk corresponding to the face - on projection of the bulge ( fig . 1 and 2 ) .
the structure of the dusty ism induces larger differences in the absolute value of the attenuation at a fixed @xmath7 for intermediate values of @xmath3 , but it does not affect the shape of the attenuation function for @xmath64 . in an analogous way , the attenuation function for a bulge associated with a dust disk having fixed opacity ( e.g. the case of @xmath18 ) becomes increasingly `` gray '' and less affected by the structure of the dusty medium when @xmath11 increases .
nevertheless , a bulge suffers less attenuation when the inclination of the galaxy approaches @xmath97 , since a larger fraction of its light is not intercepted by the dust disk .
this , of course , depends on the bulge - to - dust disk scale - height ratio , as discussed by ferrara et al .
( 1999 ) .
for the disk of a face - on galaxy , the attenuation at a fixed @xmath7 increases when @xmath3 increases , but the attenuation function does not become as gray as for the face - on bulge .
in fact the _ local _ optical depths may still be low in the outer regions of the disk .
furthermore , for the face - on disk , the absolute values and shape of the attenuation function depend more on @xmath3 than on the structure of the dusty ism ( fig . 3 and 4 ) .
also for a disk with fixed opacity ( e.g. the case of @xmath18 ) , the absolute values and shape of the attenuation function depend more on the inclination than on the structure of the dusty medium , especially for @xmath21 ( fig .
15 and 16 ) , consistent with matthews & wood ( 2001 ) .
this is at variance with the result of bianchi et al .
( 2000 ) however .
the reason for this discrepancy is the different description of the distribution and the clumpiness of the dusty ism , as discussed by the latter authors .
the previous results on the attenuation function allow us to understand the behavior of the color excess .
for a face - on bulge , we find that @xmath51 reaches an asymptotic value of @xmath113 mag already for @xmath36 , while @xmath52 increases up to 0.20.3 mag along the whole range in disk opacity probed by the models .
we remember that saturation of a color excess is reached when line - of - sight optical depths of the order of unity or more are reached for the effective wavelengths of both bands defining that color .
conversely , for the face - on disk , both @xmath51 and @xmath52 increase monotonically up to @xmath114 mag when @xmath3 increases .
the view angle of the galaxy affects the color excess , of course . for a bulge associated with a disk having @xmath18
, @xmath51 stays constant up to @xmath115 and drops almost by half for higher inclinations , while @xmath52 increases from @xmath56 up to @xmath59 mag , when @xmath11 reaches @xmath116 and drops to 0.14 mag for @xmath117 .
conversely , for a disk with @xmath18 , both @xmath51 and @xmath52 increase monotonically when @xmath11 increases . in general , a homogeneous dust distribution produces larger values of @xmath51 and @xmath52
. however , for systems seen almost edge on , the structure of the dusty ism has no impact on the color excess for either bulge or disk , consistent with the result of matthews & wood ( 2001 ) for their edge - on disk models . in general ,
the structure of the dusty ism of the disk has a larger effect on the observed magnitudes than on the observed colors of either bulge or disk component , according to our models .
the study of the behavior of the optical and near - ir color excesses of bulge and disk , as a function of disk opacity and galaxy inclination has important consequences for the interpretation of the colors of galaxies spanning the whole sequence of late hubble types .
for example , fioc & rocca - volmerange ( 1999 ) find that the observed total @xmath118 and @xmath119 colors of sa sbc galaxies are redder than those of elliptical and lenticular galaxies by 0.04 and 0.05 mag , respectively .
they comment that `` the differential internal extinction , negligible in the near - ir , is also unlikely to cause this phenomenon '' .
our previous results clearly contradict their conclusion . _
these results hint at a sensitive differential effect of the internal extinction in the near - ir , especially for the ( exponential ) bulge of a late - type galaxy_. in case of a galaxy with average inclination @xmath115 and @xmath120 , an exponential bulge may by reddened by @xmath121 mag ( fig .
19 ) against a value of @xmath122 mag for its disk ( fig .
these figures are consistent with the fact that observed , total @xmath123 colors of bulge - dominated , early - type spiral galaxies are redder than those of e / s0 galaxies by 0.09 mag , when taking into account that these galaxies have more @xmath2-like bulges ( see e.g. andredakis et al .
1995 ; graham 2001 ; hunt et al .
2004 ) .
furthermore , we find that _ a measure of the total - to - selective attenuation ratio is not equivalent to a measure of the total - to - selective extinction ratio _ , that characterizes the dust mixture present in a dusty galaxy : the former ratio changes a lot as a function of the opacity and inclination of the system .
this study focuses on the properties of the scattered light more than others , also including anisotropic multiple scattering , for bulge@xmath1disk systems .
we find , not surprisingly ( e.g. witt & gordon 1996 ) , that a homogeneous dust distribution absorbs and scatters more light at any fixed @xmath7 than a two - phase clumpy one , whether this light was produced either in the bulge or in the disk , for almost all values of @xmath3 and @xmath11 .
in fact , for the edge - on view of a spiral galaxy with @xmath18 , the fraction of light emitted by either the bulge or the disk towards the outside observer does not depend on the clumpiness of the dusty ism , whatever @xmath7 . _
the fraction of light produced either in the bulge or in the disk at a fixed @xmath7 and scattered towards the outside observer is also a function of @xmath3 and @xmath11 , and so is the color of the total scattered light_. in particular , for a given @xmath7 , this fraction increases with increasing disk - averaged optical depth along the observer s line of sight , up to the point that most of the single and multiple scattered photons get eventually absorbed or leave the system in a direction different from the observer s line of sight ( especially for optical / near - ir wavelengths ) .
dust grains scatter more efficiently and in a more forward direction , as well as absorb , the shorter the wavelength of the photons ( see table 1 ) .
thus , for a fixed inclination ( e.g. @xmath17 ) , the color of the light emitted from either the bulge or the disk and scattered towards the outside observer will turn from blue to red when @xmath3 increases .
conversely , for a fixed value of disk opacity ( e.g. in the case of @xmath18 ) , the color of the light emitted by either bulge or disk and scattered towards the outside observer will stay almost the same when @xmath11 increases ( except for the edge - on disk ) .
_ for the light received by the outside observer and produced either in the bulge or in the disk at a given wavelength , the fraction which is scattered by the dust corresponds up to 2030 per cent of the fraction which reaches the observer without being affected by the dust , for the face - on bulge , and up to 1095 per cent for the face - on disk , as a function of the opacity and the structure of the dusty ism_. for the outside observer , the ratio of scattered - to - directly transmitted light , @xmath88 , decreases ( increases ) with increasing @xmath3 for the face - on bulge ( disk ) , whatever the emission wavelength and whatever the structure of the dusty medium .
this ratio decreases with increasing inclination for the bulge of a galaxy with @xmath3 equal to unity , whatever @xmath7 and whatever the structure of the dusty ism . for the disk of the same galaxy ,
the behavior of @xmath88 as a function of @xmath11 is more complex .
this time @xmath88 increases with increasing inclination , for @xmath111 @xmath11 @xmath112 , whatever @xmath7 and whatever the structure of the dusty ism ; it decreases with increasing inclination , for @xmath124 @xmath11 @xmath125 , when @xmath126 , whatever the structure of the dusty ism ; it increases ( stays almost constant ) with increasing inclination , for @xmath124 @xmath11 @xmath125 , when @xmath127 and the structure of the dusty ism is two - phase clumpy ( homogeneous ) .
of course this result depends also on the wavelength dependence of the albedo and the scattering phase function asymmetry parameter assumed for the mixture of dust grains and on the relative distribution of photons of given wavelength and dust within the disk ( see table 1 ) .
in general , the contribution of the scattered radiation to the total monochromatic light received by the outside observer is significant , especially at uv wavelengths , even for a two - phase , clumpy , dusty ism . hence understanding dust scattering properties ( see e.g. gordon 2004 for a discussion )
is fundamental for the interpretation of e.g. the uv extragalactic observations made with _ galex _
( milliard et al . 2001 ) .
finally , our investigation confirms that _ radiative transfer effects may dilute the relative prominence of the extinction feature _ at @xmath42 , distinctive of the milky way - type dust present in our models , when the attenuation optical depth in the far - uv range is large enough , for either bulge or disk , whatever the structure of the dusty ism .
this is in agreement with the results of cimatti et al .
( 1997 ) and ferrara et al .
the latter authors attributed the disappearence of this feature to scattering modulation due to geometry and multiple scattering of photons on dust grains .
in fact the probability for events of ( multiple ) scattering increases when the disk - averaged optical depth along the observer s line of sight increases .
however , multiple scattered photons emitted at neighboring wavelengths of the @xmath42-extinction bump will be eventually absorbed , owing to the wavelength - dependence of the scattering phase function asymmetry parameter .
hence absorption will increase faster at neighboring wavelengths of @xmath42 .
the models of a bulge@xmath1disk galaxy used here simulate structural characteristics of an sbc galaxy like our own , nevertheless the results of sect .
3.1.2 and 3.2.2 may be qualitatively applied to spiral galaxies of earlier type ( cf .
ferrara et al .
1999 and tuffs et al .
2004 ) . for a quantitative conclusion , further modeling is needed , which takes into account observed scaling relations for bulge and disk components as a function of hubble type and luminosity ( e.g. andredakis et al .
1995 ; graham 2001 ; hunt et al .
2004 ) as well as the observed dependence of disk opacity on galaxy luminosity ( giovanelli et al . 1995 ; wang & heckman 1996 ; tully et al . 1998 ; masters et al .
these physical aspects are not easy to single out from observations , however , as discussed by tuffs et al .
in addition , sytematic differences in the dust / stars configurations and in the structure of the dusty ism seem to exist between giant and dwarf late - type galaxies ( dalcanton et al . 2004 ) and need to be taken into account .
we will address all these issues in future publications of this series .
this work was supported partly through grant nag 5 - 9202 from the national aeronautics and space administration to the university of toledo .
we are grateful to the referee , w. c. keel , for providing perceptive and helpful comments .
andredakis , y. c. , peletier , r. f. , & balcells , m. 1995 , , 275 , 874 audic , s. , & frisch , h. 1993 , j. quant .
transf . , 50 , 127 baes , m. , & dejonghe , h. 2001 , , 326 , 733 berlind , a. a. , quillen , a. c. , pogge , r. w. , & sellgren , k. 1997 , , 114 , 107 bianchi , s. , ferrara , a. , & giovanardi , c. 1996 , , 465 , 127 bianchi , s. , ferrara , a. , davies , j. i. , & alton , p. b. 2000 , , 311 , 601 blitz , l. 1991 , the physics of star formation and early stellar evolution , c. j. lada , n. d. kylafis , dordrecht : kluwer , 3 bruzual , a. g. , magris , g. , & calvet , n. 1988 , , 333 , 673 byun , y. i. , freeman , k. c. , & kylafis , n. d. 1994 , , 432 , 114 calzetti , d. 2001 , , 113 , 1449 calzetti , d. , bohlin , r. c. , gordon , k. d. , witt , a. n. , & bianchi , l. 1995 , , 446 , l97 charlot , s. , & fall , s. m. 2000 , , 539 , 718 cimatti , a. , bianchi , s. , ferrara , a. , & giovanardi , c. 1997 , , 290 , l43 clemens , d. p. , sanders , d. b. , & scoville , n. z. 1988 , , 327 , 129 dalcanton , j. j. , & bernstein , r. a. 2002 , , 124 , 1328 dalcanton , j. j. , yoachim , p. , & bernstein , r. a. 2004 , , 608 , 189 de grijs , r. 1998 , , 299 , 595 de grijs , r. , & peletier , r. f. 2000 , , 313 , 800 de grijs , r. , & van der kruit , p. c. 1996 , , 117 , 19 de jong , r. s. 1996 , , 313 , 377 de vaucouleurs , g. , 1948 , ann.dap . , 11 , 247 dickey , j. m. , & garwood , r. w. 1989 , , 341 , 201 disney , m. , davies , j. i. , & phillips , s. 1989 , , 239 , 939 domingue , d. l. , keel , w. c. , ryder , s. d. , white , r. e. 1999 , , 118 , 1542 draine , b. t. 2003 , , 598 , 1017 dupac , x. , et al .
2003 , , 344 , 105 emsellem , e. 1995 , , 303 , 673 ferrara , a. , bianchi , s. , cimatti , a. , & giovanardi , c. 1999 , , 123 , 437 fioc , m. , & rocca - volmerange , b. 1999 , , 351 , 869 giovanelli , r. , haynes , m. p. , salzer , j. j. , wegner , g. , da costa , l. n. , & freudling , w. 1995 , , 110 , 1059 gordon , k. d. 2004 , astrophysics of dust , a. n. witt , g. c. clayton , b. t. draine , san francisco : asp , 77 gordon , k. d. , calzetti , d. , & witt , a. n. 1997 , , 487 , 625 gordon , k. d. , misselt , k. a. , witt , a. n. , & clayton , g. c. 2001 , , 551 , 269 gordon , k. d. , clayton , g. c. , misselt , k. a. , landolt , a. u. , & wolff , m. j. 2003 , , 594 , 279 graham , a. w. 2001 , , 121 , 820 hobson , m. p. , & scheuer , p. a. g. 1993 , , 264 , 145 hunt , l. k. , pierini , d. , & giovanardi , c. 2004 , , 414 , 905 keel , w. c. , & white , r. e. 2001 , , 122 , 1369 kuchinski , l. e. , terndrup , d. m. , gordon , k. d. , & witt , a. n. 1998 , , 115 , 1438 kylafis , n. d. , & xilouris , e. m. 1996 , new extragalactic perspectives in the new south africa , d. l. block , j. m. greenberg , dordrecht : kluwer , 195 masters , k. l. , giovanelli , r. , & haynes , m. p. 2003
, , 126 , 158 matthews , l. d. , & wood , k. 2001 , , 548 , 150 mihalas , d. , & binney , j. 1981 , galactic astronomy , d. mihalas , j. binney , san francisco : freeman milliard , b. , et al .
2001 , mining the sky , a. j. banday , s. zaroubi , m. bartelmann , heidelberg : springer - verlag , 201 misiriotis , a. , & bianchi , s. 2002 , , 384 , 866 natta , a. , & panagia , n. 1984 , , 287 , 228 panuzzo , p. , bressan , a. , granato , g. l. , silva , l. , & danese , l. 2004 , , 409 , 99 pierini , d. , gordon , k. d. , & witt , a. n. 2003 , rmxac , 17 , 200 popescu , c. c. , misiriotis , a. , kylafis , n. d. , tuffs , r. j. , & fischera , j. 2000 , , 362 , 138 quillen , a. c. , frogel , j. a. , kuchinski , l. e. , & terndrup , d. m. 1995 , , 110 , 156 rix , h .- w . , & rieke , m. j. 1993 , , 418 , 123 silva , l. , granato , g. l. , bressan , a. , & danese , l. 1998 , , 509 , 103 sodroski , t. j. , odegard , n. , arendt , r. g. , dwek , e. , weiland , j. l. , hauser , m. g. , & kelsall , t. 1997 , , 480 , 173 stauffer , d. 1985 , introduction to percolation theory , london : taylor & francis tuffs , r. j. , popescu , c. c. , vlk , h. j. , kylafis , n. d. , & dopita , m. a. 2004 , , 419 , 821 tully , r. b. , pierce , m. j. , huang , j .- s . , saunders , w. , verheijen , m. a. w. , & witchalls , p. l. 1998 , , 115 , 2264 valencic , l. a. , clayton , g. c. , & gordon , k. d. 2004 , astro - ph/0408409 vogelaar , m. g. r. , wakker , b. p. , & schwarz , u. j. 1991 , fragmentation of molecular clouds and star formation , e. falgarone , f. boulanger , g. duvert , dordrecht : kluwer , 508 wang , b. , & heckman , t. m. 1995 , , 457 , 645 white , r. e. , keel , w. c. , & conselice , c. j. 2000 , , 542 , 761 whittet , d. c. b. 2003 , dust in the galactic environment , 2nd edition , d. c. b. whittet , bristol and philadelphia : iop whittet , d. c. b. , shenoy , s. s. , clayton , g. c. , & gordon , k. d. 2004 , , 602 , 291 witt , a. n. 1977 , , 35 , 1 witt , a. n. , & gordon , k. d. 1996 , , 463 , 681 witt , a. n. , & gordon , k. d. 2000 , , 528 , 799 witt , a. n. , thronson , h. a. , & capuano , j. m. 1992 , , 393 , 611 wood , k. 1997 , , 477 , l25 xilouris , e. m. , byun , y. i. , kylafis , n. d. , paleologou , e. v. , & papamastorakis , j. 1999 , , 344 , 868 rcccr 1000.0 & 0.320 & 0.800 & 5.328 & 60.0 + 1142.9 & 0.409 & 0.783 & 3.918 & 60.0 + 1285.7 & 0.481 & 0.767 & 3.182 & 65.0 + 1428.6 & 0.526 & 0.756 & 2.780 & 70.0 + 1571.4 & 0.542 & 0.745 & 2.584 & 75.0 + 1714.3 & 0.536 & 0.736 & 2.509 & 90.0 + 1857.1 & 0.503 & 0.727 & 2.561 & 110.0 + 2000.0 & 0.432 & 0.720 & 2.843 & 130.0 + 2142.9 & 0.371 & 0.712 & 3.190 & 140.0 + 2285.7 & 0.389 & 0.707 & 2.910 & 150.0 + 2428.6 & 0.437 & 0.702 & 2.472 & 160.0 + 2571.4 & 0.470 & 0.697 & 2.194 & 170.0 + 2714.3 & 0.486 & 0.691 & 2.022 & 180.0 + 2857.1 & 0.499 & 0.685 & 1.905 & 190.0 + 3000.0 & 0.506 & 0.678 & 1.818 & 200.0 + 3776.8 & 0.498 & 0.646 & 1.527 & 220.0 + 4754.7 & 0.502 & 0.624 & 1.199 & 250.0 + 5985.8 & 0.491 & 0.597 & 0.909 & 275.0 + 7535.8 & 0.481 & 0.563 & 0.667 & 300.0 + 9487.0 & 0.500 & 0.545 & 0.440 & 330.0 + 11943.5 & 0.473 & 0.533 & 0.304 & 340.0 + 15036.0 & 0.457 & 0.511 & 0.210 & 350.0 + 18929.2 & 0.448 & 0.480 & 0.145 &
360.0 + 23830.6 & 0.424 & 0.445 & 0.100 & 370.0 + 30001.0 & 0.400 & 0.420 & 0.069 & 375.0 + | we present results of new monte carlo calculations made with the dirty code of radiative transfer of stellar and scattered radiation for a dusty giant late - type galaxy like the milky way , which illustrate the effect of the attenuation of stellar light by internal dust on the integrated photometry of the individual bulge and disk components . here
we focus on the behavior of the attenuation function , the color excess , and the fraction of light scattered or directly transmitted towards the outside observer as a function of the total amount of dust and the inclination of the galaxy , and the structure of the dusty interstellar medium ( ism ) of the disk .
we confirm that dust attenuation produces qualitatively and quantitatively different effects on the integrated photometry of bulge and disk , whatever the wavelength .
in addition , we find that the structure of the dusty ism affects more sensitively the observed magnitudes than the observed colors of both bulge and disk .
finally , we show that the contribution of the scattered radiation to the total monochromatic light received by the outside observer is significant , particularly at uv wavelengths , even for a two - phase , clumpy , dusty ism .
thus understanding dust scattering properties is fundamental for the interpretation of extragalactic observations in the rest - frame uv . |
pattern formation in reaction - diffusion systems @xcite has emerged as a mathematical paradigm to understand the connection between pattern and process in natural and sociotechnical systems @xcite .
the basic mechanisms of pattern formation by local self - activation and lateral inhibition , or short - range positive feedback and long - range negative feedback @xcite are ubiquitous in ecological and biological spatial systems , from morphogenesis and developmental biology @xcite to adaptive strategies in living organisms @xcite and spatial heterogeneity in predator - prey systems @xcite .
heterogeneity and patchiness in vegetation dynamics , associated with turing patterns in vegetation dyanmics have been proposed as a connection between pattern and process in ecosystems @xcite , suggesting a link between spatial vegetation patterns and vulnerability to catastrophic shifts in water - stressed ecosystems @xcite .
the theory of non - equilibrium self - organization and turing patterns has been recently extended to network - organized natural and socio - technical systems @xcite , including complex topological structures such as multiplex @xcite , directed @xcite and cartesian product networks @xcite .
self - organization is rapidly emerging as a central paradigm to understand neural computation @xcite .
the dynamics of neuron activation , and the emergence of collective processing and activation in the brain , are often conceptualized as dynamical processes in network theory @xcite .
self - organized activation has been shown to emerge spontaneously from the heterogenous interaction among neurons @xcite , and is often described as pattern formation in two - population networks @xcite .
localization of neural activation patterns is a conceptually challenging feature in neuroscience .
cell assemblies , or small subsets of neurons that fire synchronously , are the functional unit of the cerebral cortex in the hebbian theory of mental representation and learning @xcite .
associative learning forms the basis of our current understanding of the structure and function of neural systems @xcite .
it is also the modeling paradigm for information - processing artificial neural networks @xcite .
the emergence of cell assemblies in complex neural networks is a fascinating example of pattern formation arising from the collective dynamics of interconnected units @xcite .
understanding the mechanisms leading to pattern localization remains a long - standing problem in neuroscience @xcite .
here we show that simple mechanisms of nodal interaction in heterogeneous networks allow for the emergence of robust local activation patterns through self - organization .
the simplicity and robustness of the proposed single - species pattern - forming mechanisms suggest that analogous dynamics may explain localized patterns of activity emerging in many network - organized natural and socio - technical systems .
we demonstrate that robust local , quantized activation structures emerge in the dynamics of network - organized systems , even for relatively simple dynamics . we propose a minimal - ingredients , phenomenological model of nodal excitation and interaction within a network with heterogeneous connectivity .
our goal is to demonstrate that a simple combination of local excitation of individual units , combined with generic excitatory / inhibitory interactions between connected units , leads to self - organization , and can explain the spontaneous formation of cell assemblies without the need for synaptic plasticity or reinforcement .
our model can be understood as a network analogue of the swift - hohenberg continuum model @xcite , and is able to produce a complex suite of localized patterns .
the requirements are minimal and general : simple local dynamics based on canonical activation potentials , and interactions between nodes that induce short - range anti correlation and long - range correlation in activation . because of their robustness and localization , self - organized structures may provide an encoding mechanism for information processing and computation in neural networks .
we restrict our analysis to the simplified case of symmetric networks , but our main results can be generalized to other network topologies , including directed @xcite and multiplex @xcite networks .
a node s state of activation , measured through a potential - like variable @xmath0 , is driven by local excitation dynamics and by the interaction with other nodes in the network via exchanges through the links connecting them . in dimensionless quantities , the proposed excitation - inhibition model for the evolution of potential , @xmath1 , in each node @xmath2 , is given by the model @xmath3 where @xmath4 is a dynamic forcing term , representing a double well potential , and @xmath5 is a bifurcation parameter that will be used to establish the conditions for stability and localization of the response patterns ( fig .
the currents , @xmath6 , represent the excitatory / inhibitory interactions among nodes in the network .
the structure of these nodal interactions is one of the key pattern forming mechanisms in the present model .
we consider short - range anti - correlation , and higher - order , longer - range dissipative interactions .
this two - level interaction structure , which induces anti correlation in the short range ( nearest - neighbors , or first - order connectivity ) , and long - range correlation ( second - nearest neighbors , or second - order connectivity ) is represented in fig .
mathematically , we express the integration of synaptic contributions as @xmath7 the structure of the above nodal interactions turns the dynamics into a network anologue of the swift - hohenberg continuum model , @xmath8 which is a paradigm for pattern - forming systems @xcite . the simplest form for the interaction matrices representing these correlation / anti - correlation effects (
while ensuring that the interaction fluxes conserve mass or charge ) is based on network representation of laplacian and bi - laplacian operators , @xmath9 and @xmath10 , respectively .
the network laplacian , @xmath11 , is a real , symmetric and negative semi - definite @xmath12@xmath13@xmath12 matrix , whose elements are given by @xmath14 @xcite , where @xmath15 is the adjacency matrix of the network , @xmath16 is the degree ( connectivity ) of node @xmath17 and @xmath18 is the kronecker delta .
a diffusive , fickian - type flux of the activation potential @xmath0 to node @xmath17 is expressed as @xmath19 ( see figure [ fig : fig1]a top ) .
plain waves and wavenumbers on a network topology are represented by the eigenvectors @xmath20 and the eigenvalues @xmath21 and of the laplacian matrix , which are determined by the equation @xmath22 , with @xmath23 @xcite .
all eigenvalues are real and non - positive and the eigenvectors are orthonormalized as @xmath24 , where @xmath25 .
the elements of the bi - laplacian matrix of a network can be expressed as @xmath26 where the @xmath27 matrix has information about second order nodal connectivity and takes nonzero values if node @xmath17 is two jumps away from node @xmath28 .
the operation @xmath29 models negative diffusion ( inhibition ) from the first neighbors of node @xmath17 and at the same time diffusion from its two - jump neighborhood ( see figure [ fig : fig1]a bottom ) .
the bi - laplacian , @xmath10 , has the same eigenvectors as @xmath9 ( i.e. @xmath30 ) and its eigenvalues are the square of those of @xmath9 , @xmath31 . to understand the properties and pattern - forming mechanisms in our model
, we first investigate the stability of flat states of the dynamical system : @xmath32 flat , stationary solutions @xmath33 of eq .
( 4 ) satisfy @xmath34 , where the nodal state of activation is equal for all nodes in the network , @xmath35 . for @xmath36 ,
there are three uniform solution branches given by @xmath37 and @xmath38/2 $ ] .
it is well known in one and two dimensional continuum spaces that these uniform states can become unstable and a wealth of self - organized patterns can arise @xcite . in a linear stability analysis ,
the stability of flat stationary solutions to small perturbations is determined by the eigenvalues of the laplacian and bi - laplacian matrices . introducing small perturbations , @xmath39 , to the uniform state @xmath33 , @xmath40 , the linearized version of eq .
( 4 ) takes the form @xmath41 , where @xmath42 . after expanding the perturbation @xmath43 over the set of the laplacian eigenvectors , @xmath44 , where @xmath45 is the expansion coefficient ,
the linearized equation is transformed into a set of @xmath12 independent linear equations for the different normal modes : @xmath46 where @xmath21 are the eigenvalues of the laplacian matrix .
the @xmath47-mode is unstable when re @xmath48 is positive .
instability occurs when one of the modes ( the critical mode ) begins to grow . at the instability threshold , re @xmath49 for some @xmath50 and re @xmath51 for all other modes . in figure 1b and c we summarize the linear stability analysis of the flat states of our model on a scale - free network constructed using the barabsi - albert model ( ba ) of network growth and preferential attachment @xcite .
we find that , indeed , there is a large parameter range for which the resting potential is stable .
as we demonstrate below , in the stable regime , input stimuli may trigger localized patterns of activation .
localized activation patterns are possible due to the particular structure of the model , with short- and long - range nodal interactions .
mathematically , the localized states are homoclinic orbits in the network space around the base resting state , @xmath52 .
the existence of these homoclinic orbits can be studied using the technology developed for the linear stability analysis .
since homoclinic orbits leave the flat state as we approach a small neighborhood ( cluster ) of the network , the fixed point must have both stable and unstable eigenvalues .
we linearize eq .
( 4 ) around @xmath53 and write @xmath54 , @xmath55 , arriving to the relation @xmath56 . since the laplacian eigenvalues @xmath21 are real and non - positive values , we can write them in the form @xmath57 .
if @xmath58 the topological eigenvalues of @xmath53 form a complex quartet @xmath59 . for @xmath60
they collide pairwise on the imaginary axis , and for @xmath61 they split and remain on the imaginary axis @xmath62 . for @xmath63 two of the topological eigenvalues
collide at the origin and for @xmath64 they move onto the real axis .
these results are summarized in figure [ fig : fig2]a .
the topological eigenvalues in the neighborhood of @xmath60 are characteristics of the reversible @xmath65 resonance bifurcation .
theory shows that under certain conditions the hyperbolic regime contains a large variety of topologically localized states @xcite . to understand the onset of localized patterns for different model parameters and input stimuli
, we construct the bifurcation diagram of the resting state , as a function of the total potential energy of the stimulus and bifurcation parameter @xmath5 , in the vicinity of @xmath66 @xcite .
a single bifurcation branch constructed using a pseudo - arclength continuation method @xcite has a characteristic snaking " structure of localized states with varying activation energy @xmath67 ( fig .
[ fig : fig2]b ) . as the system jumps from one steady state branch to the next one , a new neighborhood in the network is being activated .
figure [ fig : fig2]c visualizes the different steady localized states of the six different branches as they are spotted in the diagram of figure [ fig : fig2]b .
the response of the system is _ quantized _ : the transition from one pattern of activation to another one is discontinuous as we vary the activation energy @xmath68 , or the parameter @xmath5 ( fig .
[ fig : fig2]b ) .
these jumps in activation energy correspond to the addition of neighbor nodes to the cluster ( fig .
[ fig : fig2]c ) . the discontinuous
quantized nature of the network response leads to _ robustness _ in the local , final equilibrium patterns with respect to the input signal amplitude . to gain insight into the robustness of the localized patterns of activation
, we performed a synthetic test in which we initially stimulate a specific neighborhood in the network , where we set @xmath69 ( i.e. a step - like function signal in network topology ) and let the system evolve to equilibrium without decay .
we gradually increase the amplitude @xmath70 of the initial signal , and record the final energy values of the equilibrium , localized states . for small amplitudes
the perturbation relaxes back to the resting state , and no activation pattern is elicited .
there is a threshold in the energy of the input stimulus beyond which robust quantized states are form .
the states are robust in the sense that further increments in the input signal amplitude do not change the final equilibrium pattern ( fig . [
fig : fig3]a ) .
the self - organized local structures are also robust with respect to random noise in the initial stimulus .
we perform monte carlo simulations that probe the impact of the noise - to - signal ratio on the energy of the emerging quantized state .
we have confirmed that the presence of small - amplitude noise has no effect on the equilibrium states of nodal activity .
as can be expected , we do observe a departure from the energy of the base equilibrium state when the noise - to - signal ratio is sufficiently large , thereby masking the base stimulus altogether ( fig .
[ fig : fig3]b ) .
our model predicts a range of parameter values where localized states disappear , and are replaced by _ global activation patterns_. mathematically , global patterns are possible when the non - active stationary solution is perturbed outside the parameter region of localized patterns ( @xmath61 ) .
these global turing patterns @xcite can be understood and modeled using the mean - field approximation ( mfa ) , a method that segregates nodes according to their degree and has been successfully used to approximate a wide variety of dynamical processes in heterogeneous networks , like epidemic spreading @xcite , activator - inhibitor models @xcite and voter models @xcite .
this theory allow us to reduce the problem to a single equation for the membrane potential for all the nodes in the system .
since in our model both the degree and two - jump degree play important role in the formation of patterns , we use a mfa where we assume that all the nodes with the same degree and two - jump degree behave in the same way .
we start by writing eq .
( 3 ) in the form @xmath71 where the local fields felt by each node , @xmath72 , @xmath73 and @xmath74 are introduced . these local fields
are then approximated as @xmath75 , @xmath76 and @xmath77 , where @xmath78 is the degree and @xmath79 is the number of secondary connections of node @xmath17 ( two - jump degree ) .
the global mean fields are defined by @xmath80 where @xmath81 and @xmath82 , where @xmath83 . here
, @xmath84 is the number of nodes with degree @xmath85 , @xmath86 is the number of nodes with @xmath87 number of two - jump neighbors and @xmath88 is the size of the network . in the above expressions , with @xmath89 we denote the sum over the nodes with degree @xmath85 and with @xmath90 the sum over the nodes with two - jump nodal connectivity @xmath91 . with this approximation ,
the individual model equation on each node interacts only with the global mean fields @xmath92 and @xmath93 and its dynamics is described by : @xmath94.\end{aligned}\ ] ] since all nodes obey the same equation , we have dropped the index @xmath17 and introduced the parameters @xmath95 and @xmath96 .
the activation potential depends now on the global fields @xmath92 and @xmath93 as well as on the parameter compination ( @xmath97 )
, i.e. @xmath98 . if the global mean fields @xmath92 and @xmath93 are given , the combination ( @xmath97 ) plays the role of a bifurcation parameter that controls the dynamics of each node in the system . the time independent version of above mean field equation can be written as a third degree algebraic equation that we solve @xmath12 times for the @xmath12 nodes in the system .
for each node @xmath17 , we get three solutions @xmath99 , @xmath100 that can be stable or unstable depending on the sign ( negative or positive respectively ) of the operator @xmath101 . after tuning the bifurcation parameter @xmath5 to a negative value
, we can compute the global turing pattern from direct numerical simulations and determined the global mean fields @xmath92 and @xmath93 .
each node @xmath17 in the network is characterized by its degree and second nodal connectivity , so that it possesses a certain parameter combination , ( @xmath97 ) . substituting
these computed global mean fields as well as the values of @xmath47 and @xmath102 into equation ( 7 ) , bifurcation diagrams of a single node can be obtained and projected onto the turing pattern . in figure
( 4a ) we show for our toy network model " that the stable brunches of the nodal bifurcation diagrams calculated using the mfa fit very well the computed turing pattern .
we further assess the dependence of the network topology on the global pattern formation and we find that when the degree distribution is narrower compared to a scale - free network , the distribution of the ( @xmath97 ) is more homogeneous and therefore the stationary turing patterns look smoother ( figure 4b and c ) .
therefore , global network turing patterns are essentially explained by the bifurcation diagrams of individual nodes coupled to the global mean fields , with the coupling strength determined by their degree and two - jump connectivity .
our results suggest a new mechanism for the formation of localized nodal assemblies in networks , arising from long - range second neighbor interactions . rather than relying on reinforcing mechanisms
synaptic plasticity , we show that localized , robust nodal assemblies are possible due to self - organization . the emergence of localized activation patterns derived from the simple and general functional structure of our proposed conceptual model : local dynamics based on activation potentials , and interactions between nodes that induce short - range anticorrelation and long - range correlation in node - to - node exchanges .
the proposed system is a network analogue of the swift - hohenberg continuum model , and is able to produce a complex suite of robust , localized patterns .
hence , the spontaneous formation of robust operational cell assemblies in complex networks can be explained as the result of self - organization , even in the absence of synaptic reinforcements .
hence , these self - organized , local structures can provide robust functional units to understand natural and technical network - organized processes .
funding for this work was provided by an mit vergottis graduate fellowship and a mcdonnell postdoctoral fellowship in studying complex systems ( to cn ) , the us department of energy through a doe career award ( grant de - sc0003907 ) and a doe mathematical multifaceted integrated capability center ( grant de - sc0009286 ) ( to rj ) , and a ramn y cajal fellowship from the spanish ministry of economy and competitiveness ( to lcf ) .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . , , , &
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , &
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , &
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , &
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . &
. _ _ * * , ( ) . .
_ _ * * , ( ) . .
_ _ * * , ( ) . , , , &
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
cn , rj and lcf designed the research , performed the analysis and wrote the manuscript .
the authors declare no competing financial interests . . in the inset
we show the potential landscape minus the integral of @xmath103 with respect to u , which exhibits a single well ( at @xmath53 ) with an inflection point , a necessary condition for localized patterns to exist .
nodes interact in the network through diffusion - like exchanges via the links connecting them .
the network laplacian operator , @xmath104 , represents short range diffusion of the species in the system ( top ) .
the network bi - laplacian operator , @xmath105 , induces short range anti - correlation with the nearest - neighbors , and long - range correlation with the second - nearest neighbors ( bottom ) . *
( b c ) * _ linear stability analysis of the flat stationary solutions of our model . _ * ( b ) * the maximum value of the growth rate @xmath106 as a function of the bifurcation parameter @xmath5 for the two flat stationary states @xmath107 ( brown ) and @xmath108 ( blue ) on a barabsi - albert network model with mean degree @xmath109 and size @xmath12=2000 .
when the maximum value of @xmath106 is negative , the state is stable with respect to small non uniform perturbation .
( inset ) the growth rate @xmath106 as a function of the laplacian eigenvalue @xmath110 ( eq .
( 5 ) ) for three different values of the bifurcation parameter @xmath5 as they indicated in the main diagram for the flat stationary solution @xmath108 . *
( c ) * the flat stationary solutions @xmath111 and @xmath112 as a function of @xmath5 on the same network .
solid ( dotted ) lines represent stability ( instability ) with respect to small non - uniform perturbations .
the labelled bifurcation points are @xmath113 , @xmath114 and @xmath115 and @xmath116 .
the pink shaded region is where we observe localized self - organization patterns with respect to the trivial solution @xmath111 . for values of
@xmath5 outside that region we get either global activation patterns ( for @xmath117 ) or any perturbation relaxes back to the flat stationary solution ( for @xmath118).,width=552 ] .
for positive values of @xmath5 the trivial stationary solution is stable with respect to uniform small random perturbations ( solid line ) while for negative values of @xmath5 this state becomes unstable ( dotted line ) .
also shown in the insets are the topological eigenvalues of the trivial state as we tune the bifurcation parameter .
the behavior of eigenvalues in the neighborhood of @xmath60 indicates the possibility for localized patterns in the neighborhood of small positive values of @xmath5 ( pink shaded region ) . *
( b ) * a single branch of the bifurcation diagram in a barabsi - albert network model of size @xmath12=200 with mean degree equal to @xmath119 and minimum degree equal to 1 .
solid ( dotted ) lines represent stable ( unstable ) localized solutions . *
( c ) * visualization of the localized patterns corresponding to the states indicated on the bifurcation diagram ( b ) .
gray - colored nodes are non - active ( @xmath53 ) , red - colored nodes are active with @xmath120 and blue - colored nodes are active with @xmath121 .
the size of the node is proportional to its eigenvalue centrality.,width=600 ] at the nearest and next - nearest neighbors of the best connected node in the system .
when the amplitude is very small , the initial perturbation relaxes back to the trivial solution and no quantized state is formed ( _ i _ ) . as the amplitude of the input signal is increased , fragile quantized states are formed ( _ ii _ ) . when the amplitude of the input signal is larger than a threshold value , a very robust quantized state is formed ( _ iii _ ) .
further increases in the input signal amplitude lead to the same quantized state .
( insets ) visualization of the input signal in our network topology ( the amplitude increases from left to right ) as well as the resulting equilibrium state . *
( b ) * the energy of the resulting quantized state with respect to the ratio between the signal amplitude and the noise amplitude . starting from the step - like input signal that gives the robust quantized state , we add random noise at the already perturbed neighborhood and we compute the energy of the resulting quantized state over 100 realizations .
we use a barabsi - albert scale - free network of size @xmath12=200 and mean degree @xmath122=4 .
, width=504 ] ) .
the initial exponential growth of the perturbation is followed by a nonlinear process leading to the formation of stationary turing patterns .
* ( a ) * ( left ) the activation profile as a function of the node index @xmath17 of a global stationary turing pattern from direct simulation ( blue crosses ) is compared with the mean - field bifurcation diagram .
black curves indicate stable branches while grey curves correspond to unstable branches of a single activator inhibitor system coupled to the computed global mean fields .
we sort the node index @xmath17 in increasing connectivity @xmath85 .
nodes with the same degree are sorted with increasing two - jump connectivity @xmath87 ( see inset ) .
we use the same barabsi - albert network model as in fig . 2 and we set the bifurcation parameter equal to @xmath123 .
we have confirmed that similar results hold for larger network sizes .
( right ) visualization of the global activity pattern on the network topology . *
( b ) * the activation profile as a function of the node index @xmath17 of global stationary turing patterns from direct simulation for bifurcation parameter @xmath124 on an erds - rnyi random network with size @xmath125 and mean degree @xmath126 ( blue curve ) along with the stable branch of the mean field approximation ( black curve ) . * ( c ) * the activation profile as a function of the node index @xmath17 of global stationary turing patterns from direct simulation on a barabsi - albert scale free network with the same mean degree and the same number of node as the erds - rnyi of b. we sort the node index in increasing connectivity @xmath85 and two - jump connectivity @xmath87 .
, width=456 ] | self - organization and pattern formation in network - organized systems emerges from the collective activation and interaction of many interconnected units .
a striking feature of these non - equilibrium structures is that they are often localized and robust : only a small subset of the nodes , or cell assembly , is activated .
understanding the role of cell assemblies as basic functional units in neural networks and socio - technical systems emerges as a fundamental challenge in network theory .
a key open question is how these elementary building blocks emerge , and how they operate , linking structure and function in complex networks . here
we show that a network analogue of the swift - hohenberg continuum model a minimal - ingredients model of nodal activation and interaction within a complex network is able to produce a complex suite of localized patterns .
hence , the spontaneous formation of robust operational cell assemblies in complex networks can be explained as the result of self - organization , even in the absence of synaptic reinforcements .
our results show that these self - organized , local structures can provide robust functional units to understand natural and socio - technical network - organized processes . |
chandra observatory x - ray images of the hot atmospheres of galaxy clusters show a wealth of structure associated with central radio sources , including cavities , shock fronts , metal - enriched plumes , and filaments ( see peterson & fabian 2006 and mcnamara & nulsen 2007 for reviews ) .
it was realized early on that measurements of cavity sizes and their surrounding pressures provide a gauge of the @xmath10 work ( mechanical energy ) expended by radio jets as they inflate cavities against the surrounding gas pressure ( mcnamara et al .
2000 ) . combining this with the assumption that cavities are driven to their current locations primarily by buoyant forces , the derived mechanical energy and mean jet power lie in the ranges @xmath11 erg and @xmath12 , respectively .
systematic studies of clusters and giant elliptical galaxies have shown that mean jet power estimated in this way is comparable to the power required to quench cooling flows ( brzan et al .
2004 , dunn & fabian 2006 , rafferty et al .
2006 , nulsen et al .
2007 , diehl & statler 2008 , diehl et al . 2008 ) .
most of this energy is expected to heat the surrounding gas ( churazov et al .
2001 , 2002 , brggen & kaiser 2001 , ruszkowski et al .
2004 , mccarthy et al .
2004 , heinz & churazov 2005 , voit & donahue 2005 ) . observations and simulations imply that massive black holes located in the nuclei of elliptical and brightest cluster galaxies ( hereafter bcgs ) combined with an abundance of circumnuclear fuel draining onto them provide a natural feedback mechanism that is able to maintain most of the cooling gas at x - ray temperatures ( pizzolato & soker 2005 , sijacki et al .
these developments have significant implications for the formation and evolution of galaxies and the fundamental properties of radio jets .
accretion - driven outflows may have regulated the growth of bulges giving rise to the correlation between bulge mass and smbh mass ( ferrarese & merritt 2000 , gebhardt et al .
2000 , tremaine et al .
2002 , hring & rix 2004 , di matteo et al .
agn feedback at late times in the so - called `` radio mode , '' may be responsible for the turnover at the bright end of the galaxy luminosity function and the dearth of bright blue bulge galaxies at late times ( bower et al .
2006 , croton et al . 2006 , sijacki et al . 2007 , 2008 , somerville et al . 2008 ) .
agn feedback may explain other significant problems including the breaking of self - similarity of the cluster scaling relations ( borgani et al . 2005 , puchwein et al .
2008 , gitti et al .
2007 , cavaliere et al . 2002 ) , variations in the baryon fraction in clusters ( allen et al .
2004 , vikhlinin et al .
2006 , gitti et al .
2007 ) , and the production of energetic particles ( benford & protheroe 2008 ) . furthermore , because the @xmath10 methodology gives a fairly reliable lower limit to the the enthalpy , or total jet energy ( jones & de young 2005 , binney et al .
2007 , nusser , silk , & babul 2006 , mathews & brighenti 2008 ) , it is able to levy interesting constraints on the particle and magnetic field content , and radiative efficiencies of extragalactic radio sources ( brzan et al . 2004 , 2008 , de young 2006 , dunn , fabian , & celotti 2006 , croston et al .
2008 , diehl et al .
2008 , li et al .
2006 , nakamura et al .
2007 , 2008 ) . here
we show that this methodology is able to place interesting constraints on the mass and spin of black holes driving agn activity .
agn powering mechanisms can be broadly described in the context of the accretion and spin paradigms ( see meier 2002 for a review ) . in the accretion paradigm
, agn outflows are powered by gravitational binding energy released by infalling gas ( eg . , begelman , blandford & rees 1984 ) .
the inflow of gas onto a disk eventually drains onto the black hole as its angular momentum is transported outward through shear stresses ( shakura & sunyaev 1973 ) , likely dominated by magnetohydrodynamic turbulence ( eg . ,
balbus & hawley 1991 ) . when the accretion rate approaches the eddington value , @xmath13 , an optically thick disk forms releasing binding energy in the forms of radiation and winds .
in the low accretion limit , ie . , @xmath14 , a geometrically thick , optically thin disk forms .
most of the energy dissipated in the disks of these systems is locked - up in inefficiently radiating ions that are carried inward in an advection dominated accretion flow ( adaf ; narayan & yi 1995 ) .
some or most of the binding energy is released in a mechanical outflow associated with a radio jet .
alternatively , it is possible that much of the matter entering the accretion disk never reaches the black hole , but instead is blown away in a wind , such as the adios model of blandford & begelman ( 1999 ) .
adaf and adios models and their kin may be broadly applicable to black holes accreting far below the eddington rate in cooling atmospheres of clusters and elliptical galaxies .
the spin paradigm posits that the rotational energy of a rapidly spinning black hole and its accompanying disk is channeled outward through the formation of a magnetohydrodynamical jet .
jet power may be generated by rotation of the accretion disk ( blandford & payne 1982 , bp ) , or by rotation of the black hole and its ergosphere ( blandford & znajek 1977 , bz ) , or a combination of both ( meier 1999 , nemmen et al .
2007 ) . like the accretion model itself ,
spin - powered jets are maintained by accretion of a weakly magnetized plasma ( meier 1999 ) .
however , the power generated per unit of accreted mass can , in some models , be much larger than for purely accretion powered models .
the existence of a population of spin - powered radio galaxies was inferred by sikora et al .
( 2007 ) who found two distinct loci in the eddington accretion luminosity versus radio power plane . in this plane ,
radio power declines with increasing nuclear accretion luminosity in both populations when expressed in eddington units ( see also ho 2002 ) .
the lower radio power population is composed primarily of disk galaxies while the higher power population is composed primarily of elliptical galaxies .
sikora et al . and
others have argued that this bimodality indicates two processes are powering radio galaxies .
the lower - power disk systems , which contain ample levels of cold gas , have slowly rotating holes that are powered primarily by accretion .
higher radio power elliptical galaxies , which are nearly devoid of cold gas , harbor rapidly spinning black holes that are able to power their radio jets .
sikora et al .
pointed out that unlike disk galaxies , the hierarchical growth of elliptical galaxies will lead to black hole mergers that , under the right conditions , spin - up smbhs , providing a natural mechanism to power radio jets ( see also meier 2002 , but see hughes & blandford 2003 ) .
earlier studies have attempted to evaluate the viability of spin powering of extragalactic radio jets .
for example , cao & rawlings ( 2004 ) used 3cr radio galaxies to constrain bz and bp models , while nemmen et al .
( 2007 ) evaluated the performance of a hybrid model incorporating elements of the bz , bp , and adaf models .
these studies concluded that the bz model was unable to power the most luminous sources , but that hybrid models ( eg . , meier 1999 , punsly & coroniti 1990 ) may be able to do so .
bcgs centered in cooling flows are often rich in atomic and molecular gas ( edge 2001 , salome & combes 2003 , donahue et al .
2000 ) and they frequently have an ample supply of fuel to power agn by accretion .
when this is not true , as we argue for ms0735.6 + 7421 ( ms0735 ) , black hole spin provides a plausible alternative power supply .
the energy available in a maximally spinning , @xmath7 black hole @xmath15 is comparable to the x - ray luminosity of a cooling flow integrated over the ages of clusters .
therefore , spin may be an energetically significant factor in the evolution of cooling flows , and in the creation of excess energy ( entropy ) in groups and clusters ( voit 2005 , babul et al .
2002 , borgani et al .
2005 ) . in this paper , we focus on the the ms0735 cluster , which harbors an energetic agn outburst that was identified by its unusually large pair of x - ray cavities ( mcnamara et al .
each is roughly 200 kpc in diameter and is filled with radio emission ( fig .
the cavities are surrounded by a weak but powerful shock front reaching beyond the central galaxy into the inner several hundred kiloparsecs of the cluster . with a total energy of @xmath16 erg and a mean jet power exceeding @xmath17 ,
its prodigious power demands push accretion and jet models to their limits .
we show that both the energy and power output from agn determined using x - ray cavities and shock fronts provide interesting constraints on black hole mass , spin and accretion power .
the accretion model requires an implausibly large accretion rate compared to the available fuel supply and rate of star formation in ms0735 s bcg , unless the smbh is unusually massive .
however , the rotational energy of a rapidly spinning smbh with a mass consistent with the magorrian relation can accommodate its demanding jet energy and power output .
our case for spin powering is based on straightforward energy and gas supply arguments that avoid environmental complexities , orientation effects ( urry & padovani 1995 , antonucci 1993 ) , and the details of spin models ( eg . , cao & rawlings 2004 , nemmen et al .
2007 , reynolds et al .
we attempt to balance our discussion by pointing out problems with the spin paradigm as it applies to the quenching of star formation in massive galaxies at late times . throughout this paper
we assume a @xmath18 cosmology with @xmath19 and @xmath20 for redshift @xmath21 , the corresponding ratio of linear to angular size is @xmath22 .
an image of the inner 200 arcsec ( 700 kpc ) of the cluster combining the x - ray , @xmath23-band , and radio wavelengths is shown in fig .
the single orbit image was exposed with the hubble space telescope s advanced camera for surveys ( acs ) through the f850lp filter , which we refer to as the @xmath23 image .
its angular resolution is @xmath24 arcsec per pixel .
individual frames were reduced and calibrated using standard pipeline processing and were combined using the multi - drizzle technique .
magnitudes quoted here were corrected for foreground galactic extinction , k - correction , evolution of the stellar population , cosmological surface brightness dimming , and an aperture correction .
these corrections are given in table 1 .
further analysis was performed using the iraf and idl image processing environments . the 320 mhz radio image and 40 ksec chandra x - ray image shown in fig .
1 are discussed in brzan et al .
( 2008 ) and mcnamara et al .
( 2005 ) , respectively .
a new h@xmath25 image was obtained on 22 january , 2005 with the wiyn 3.5 m telescope on kitt peak using the mosaic camera .
individual frames were taken through the narrow , on - band filter w029 and off - band gunn r ( w009 ) filter exposed for 10,800 sec and 3600 sec , respectively .
the images were corrected for bias and flat fielding using standard procedures .
the star formation rate of the bcg was measured using a uv image exposed with xmm - newton s optical monitor wide 2 ( w2 ) imager for 18.5 ksec .
the image was obtained in tandem with a deep x - ray observation of ms0735 presented in gitti et al .
the w2 filter has a passband spanning @xmath26 , and the camera provides a spatial resolution of @xmath27 arcsec fwhm .
fig . 1 shows radio jets emerging from the nucleus at an angle of roughly 45 degrees , to the north - east and south - west .
the jets are redirected to the north and south at distances of 5090 kpc from the nucleus , where they expand into lobes .
the radio lobes are enclosed by two x - ray cavities each of which is nearly 200 kpc ( 1 arcmin ) in diameter .
the lobes have displaced roughly a trillion solar masses of x - ray plasma .
the cavities are surrounded by a weak but powerful shock front with mach number of 1.4 measured to the east and west of the agn , and perpendicular to the radio axis ( mcnamara et al . 2005 ) .
a spherical model for this part of the shock requires a driving energy of @xmath28 erg .
this approach underestimates the true energy because the shock front extends considerably further to the north and south , implying a faster , stronger shock , encompassing a greater volume in those directions .
the age of the model shock , @xmath29 yr , is better determined , since it depends largely on the shock radius and its current speed .
assuming @xmath30 per cavity , rafferty et al .
( 2006 ) found the total enthalpy of the cavities to be @xmath31 erg , which is close to the shock energy .
a faithful outburst model would account for the entire shock front and the cavities .
in particular , it should explain the division of outburst energy between the shock flow and cavity enthalpy .
without such a model , and given the similarity of the two energy estimates , and that the shock energy is probably underestimated , it is reasonable to compute the total energy of the outburst as their sum , @xmath32 erg .
the mean power of the jet is then @xmath33 . assuming the outburst was powered by the gravitational binding energy released by accretion , and adopting a mass - energy conversion efficiency @xmath34
, we find that under these assumptions the black hole grew by @xmath35 here , @xmath36 accounts for the lost binding energy , and @xmath37 is the total energy output in mechanical and radiative forms .
we have ignored radiation because it accounts for a negligible fraction of the current power output .
this growth in mass corresponds to an average growth rate of @xmath38 over the past @xmath39 .
rafferty et al ( 2006 ) found a black hole mass of @xmath40 using the bcg s 2mass k - band bulge luminosity @xmath41 and the scaling relation of marconi & hunt ( 2003 ) . applying the v - band bulge luminosity @xmath42 ( see below ) to lauer s ( 2007 ) black hole mass
versus bulge luminosity relation gives a somewhat higher mass of @xmath43 .
lauer et al .
( 2007 ) have shown that 2mass measurements systematically underestimate the true bulge luminosity .
therefore , we adopt the lauer et al .
value as the nominal progenitor mass .
taken at face value , the accretion model then implies that the smbh grew by about @xmath44 during the outburst .
the corresponding eddington accretion rate for a @xmath43 black hole is @xmath45 . powering the outburst by accretion implies that ms0735 s smbh has grown , on average , at @xmath46 of the eddington rate for @xmath47
these figures are merely indicative given the crude estimate of the black hole mass and the poorly known value of the mass - to - energy conversion factor , @xmath48 , which depends on , among other things , the spin of the black hole .
high resolution structural studies of bcgs ( e.g. , laine et al .
2003 ) have found it useful to characterize the light profile using the so - called nuker law : @xmath49^{\alpha})^{(\gamma -\beta)/\alpha}.\ ] ] a plot of the nuclear surface brightness profile of the bcg is shown in fig .
2 with the nuker law fit superposed .
profile parameters are given in table 1 .
the profile reveals no excess emission in the nucleus associated with a bright agn , nuclear star formation , or a stellar disk or cusp .
we find a break radius of @xmath50 .
the systematic error is larger than the statistical uncertainty , and depends on the asymptotic outer slope , @xmath51 , whose value is sensitive to the background correction , and to dust extinction , which extends beyond the break radius .
our attempt to model these effects suggest errors less than two or three tenths of an arcsec .
the break radius is similar in angular size to values in laine s sample of bcgs .
however , its linear size is 2 to 3 times larger than the largest in the laine et al .
( 2003 ) and lauer et al .
( 2007 ) samples .
laine et al . found a correlation between break radius and v - band galaxy luminosity ( see also faber et al .
1997 , lauer et al .
ms0735 s r - band absolute magnitude @xmath52 ( rafferty et al .
2006 ) adjusted to the v - band assuming @xmath53 is @xmath42 . a comparison between this value and laine s sample shows that ms0735 s bcg would be among the most luminous in their sample .
in fact , ms0735 s core is comparable to that of a galaxy one to two magnitudes brighter .
the implication is that it probably harbors an ultramassive black hole ( cf .
, section 4.6 ) .
the runs of ellipticity ( @xmath48 ) and position angle ( pa ) of the @xmath23-band stellar isophotes are shown in fig .
3 . beyond 1 arcsec ,
pa is roughly constant with radius approaching a value of @xmath54 degrees . over the same region ,
@xmath48 rises gradually from @xmath55 to roughly @xmath56 .
variations in @xmath48 and pa within one arcsec are primarily due to dust .
the galaxy shows no obvious evidence of other dynamical disturbances that might have been induced by a recent merger or strong gravitational interaction .
no evidence for unresolved nuclear emission is seen in fig .
we find an upper limit of @xmath57 at the resolution limit @xmath24 arcsec .
similarly , no evidence of nuclear emission is seen in the x - ray or ultraviolet bands .
an upper limit to the x - ray flux from a nuclear source was found by extracting a spectrum from the chandra image within a circular aperture 1.5 arcsec in radius located at the position of the bcg s nucleus .
only 392 net counts were extracted .
assuming the counts represent a combination of thermal x - rays from the hot gas and non - thermal x - rays from an agn , we fit the spectrum to a thermal model with temperature , foreground column density , and metallicity , held constant , while solving for an embedded power law spectrum .
the power law fit was largely unconstrained and offered no improvement over the thermal model .
the upper limit to the unabsorbed nuclear flux @xmath58 corresponds to a bolometric x - ray luminosity @xmath59 .
the uv image discussed later gives an upper limit of @xmath60 from the nucleus .
thin disk accretion models predict that most of the gravitational binding energy released by accretion should be radiated throughout the x - ray , uv , and optical bands .
the optical luminosity of such a disk for a reasonable viscosity parameter , @xmath61 , is @xmath62 ( meier 2002 ) . for ms0735
this corresponds to an optical luminosity of @xmath63 for an assumed smbh mass of @xmath64 and an accretion rate in eddington units of @xmath65 .
this figure is vastly larger than our upper limits , suggesting the emission is beamed out of our line of sight , is obscured , or is absent altogether .
the implied level of @xmath23-band extinction is 10.3 mag , corresponding to column density @xmath66 , which is normal for agn ( cf .
, risaliti et al . 1999 ) .
the absence of penetrating x - ray emission argues against a hidden nucleus .
it is possible that the energy has been advected inward by an adaf - like process ( narayan & yi 1995 ) , but this seems unlikely unless the black hole is much more massive than the bulge light to black hole mass scaling relations imply . the @xmath23-band hst image of the central galaxy is shown in fig . 4 next to the difference between the image and a smooth stellar background model
the nucleus is undistinguished .
a system of filamentary dust features is seen in the inner five arcsec ( 17 kpc ) of both the direct and difference images .
these features are labeled in fig .
5 and their properties are listed in table 2 .
no evidence for a dusty disk is seen .
laine et al . ( 2003 ) detected dust features in @xmath67 of their sample of bcgs .
nuclear dust disks were found in @xmath68 and filamentary dust features in @xmath69 of their sample . therefore
, dust is a common feature of these systems . nuclear dust and ionized gas disks several hundred pc in diameter
are often found in ge galaxies ( e.g. , ferrarese & ford 1999 ) .
such a disk would be unresolved in ms0735 .
the level of extinction can be characterized by comparing the light decrement in the subtracted image to that in the model .
assuming for simplicity that the dust is located in a foreground screen , the light decrement can be expressed as @xmath70 , where @xmath71 is the optical depth , @xmath72 is the observed surface brightness and @xmath73 is the model . further assuming a standard galactic extinction curve ( cardelli et al .
1989 ) and a mean gas to dust ratio of 100 , we find a relationship between the hydrogen column density @xmath74 toward ms0735 and @xmath75 to be @xmath76 .
@xmath75 is typically a few percent in the @xmath23 band .
the gas masses toward each feature were found by multiplying the column densities by the area subtended by the dust features .
5 highlights the deepest dust features whose total gas masses sum to @xmath77 .
the integrated absorption throughout the h@xmath25 clouds indicated by region i corresponds to a total gas mass of @xmath78 . because the foreground screen model tends to underestimate the dust mass
, the actual mass could be several times larger .
salome and combes ( 2008 ) recently found an upper limit of @xmath79 to the molecular gas mass near the nucleus of ms0735 using the iram telescope . furthermore , the mid infrared rotational lines of warm @xmath80 are weak in the spectrum of ms0735 [ zwcl 1370 ] ( de messires et al .
2009 ) , and correspond to @xmath81 of gas at temperatures between @xmath82 k. both observations are consistent with the modest gas mass implied by the dust map .
the bgc harbors a luminous h@xmath25 nebula discovered by donahue , stocke , & gioia ( 1992 ) . our image of the nebula
is shown in fig .
6 superposed against a ground - based @xmath83-band image obtained recently with the wiyn telescope .
the h@xmath25 isophotes show modest variations in ellipticity and position angle ( fig . 7 ) . a comparison between the h@xmath25 isophotes ( fig . 7 ) and the stellar isophotes ( fig .
3 ) shows that the stars and gas have similar shapes .
the gas is relaxed and has settled into the galaxy s potential well .
the only indications of an interaction are a shallow trough in the h@xmath25 emission extending north - east of the nucleus roughly following the radio jet shown in fig . 1 , and a low surface brightness h@xmath25 cloud @xmath84 arcsec to the north - east of the nucleus .
the trough may be related to a dust filament at that location .
otherwise the gas appears surprisingly placid considering its location in a galaxy that has recently experienced such a powerful agn outburst .
donahue , stocke , & gioia ( 1992 ) found an h@xmath25 + [ nii ] flux of @xmath85 , which corresponds to a luminosity of @xmath86 } = 1.22 \times 10^{42}\rm ~erg~s^{-1}$ ] . in cooling flow systems
the ratio of [ nii ] to h@xmath25 flux shows a large spread centered around unity ( hatch et al .
2007 ) . assuming the h@xmath25 luminosity is half the h@xmath25 + [ nii ] luminosity , an electron density @xmath87 , and a nebular temperature of @xmath88 k
, we find an ionized gas mass of @xmath89 .
this value is much less than the dust estimates and salome & combes upper limit , but is larger than the warm molecular hydrogen mass found by de messires et al .
@xmath90 is likely to be the mass of gas in the ionized skins of cold molecular clouds , and thus represents a small fraction of the gas present in the galaxy .
all taken together , our results are consistent with less than @xmath7 of cold gas in the inner 20 kpc of the galaxy . despite the existence of some cold gas , strong star formation is not seen .
rafferty et al .
( 2008 ) have shown that the bcg s central colors and halo color gradient are consistent with a normal old stellar population .
the color gradient lacks a prominent central blue dip seen in star forming bcgs , although we can not exclude the possibility of a modest blueing of less than @xmath91 mag in the inner 2 arcsec .
we estimate the star formation rate using the xmm - newton optical monitor wide 2 ( w2 ) uv image .
the w2 filter has a passband of @xmath26 , and the camera provides a spatial resolution of @xmath27 arcsec fwhm . the image presented in fig .
8 shows that the galaxy is barely detected in the far uv .
we find a uv luminosity of @xmath92 within a 10 arcsec aperture .
the relationship between uv luminosity and star formation rate ( salim et al . 2007 ) gives a star formation rate of @xmath93 .
both the flux and star formation rate have been k - corrected and corrected for foreground extinction .
it is difficult to disentangle the uv flux emerging from the old stellar population from that which is genuinely associated with young stars .
therefore , we interpret this measurement as an upper limit to the star formation rate .
ms0735 does not exhibit the mid - infrared cool dust continuum that normally rises from @xmath94 to the far infrared in star - forming galaxies ( de messires et al .
a limit to the star formation rate based on the @xmath95 continuum is about @xmath96 ( donahue et al .
2009 , in preparation ) .
star formation at this level should have been detected easily in rafferty s @xmath97-band observations and in our far uv data , unless it is buried in dust .
this possibility seems unlikely given the modest dust levels seen in the hst image , but we can not rule it out .
we note that the ir star formation rate found by donahue et al .
( 2009 ) assumes the dust emission is powered by young stars .
if other heat sources are operating , the ir measurement would over - estimate the star formation rate .
we adopt a conservative upper limit to the star formation rate of @xmath3 .
the upshot is that the star formation rate is modest compared to the apparent accretion rate onto the smbh .
the conditions in ms0735 stretch the ability of the various accretion mechanisms to power its agn .
these mechanisms include bondi accretion from the hot atmosphere , cold accretion from a donor galaxy , cooled gas from the hot atmosphere , stars plunging into the smbh , or inspiral of one or more smbhs from a galaxy or group merger .
each will be discussed in turn .
accretion of gas from the hot halo by the bondi mechanism is attractive for several reasons .
it is in principle straightforward to regulate in the context of feedback models and cooling flows ( eg . , nulsen & fabian 2000 , sijacki et al .
2007 , somerville et al . 2008 ) , and the x - ray atmosphere provides a steady supply of fuel . in relatively low power giant ellipticals , bondi accretion has been shown to be energetically feasible in the sense that hot atmospheres probably have a sufficient gas density to supply the mass required to account for the observed jet powers ( di matteo et al .
2000 , allen et al . 2006 , rafferty et al .
2006 ) .
the average gas density and temperature in the inner 3 arcsec ( 10 kpc ) of ms0735 s hot halo is @xmath98 @xmath99 , and @xmath100 kev , respectively ( rafferty et al .
2006 , 2008 ) .
a reanalysis of the chandra image pushing further into the center yields a temperature of @xmath101 kev and @xmath102 @xmath99 in the inner arcsec or so . adopting a nominal black hole mass of @xmath103 in units of @xmath104
, we find a bondi accretion rate of @xmath105 this value lies far below @xmath106 required to power the outburst
. the density and temperature of the hot halo may be somewhat higher and lower , respectively below chandra s resolution limit and approaching the vicinity of the black hole .
however , the combination of gas density , temperature , and black hole mass that is able to yield a sufficiently large accretion rate is extreme .
for example , by increasing the gas density to 1 particle @xmath107 and decreasing the gas temperature to 0.14 kev or so while holding the black hole mass at its nominal value would result in a sufficiently high bondi rate .
however , the volume containing the @xmath108 of gas required to fuel the outburst would be roughly 1.8 kpc ( 0.5 arcsec ) in radius .
this gas would be revealed as a sharp spike in the x - ray emission near the nucleus , which is not observed .
additional sources of uncertainty include the poorly constrained energy conversion efficiency which may fall below the canonical 10% value , as suggested by allen et al .
( 2006 ) and merloni & heinz ( 2007 ) .
conversely , the efficiency may be larger than @xmath44 if the smbh is indeed rapidly spinning . while we can not rule out bondi accretion based on these arguments , we regard it as exceedingly unlikely if the smbh mass lies close to our adopted value .
the demands on gas density and temperature are eased dramatically for a black hole mass substantially in excess of @xmath109 .
a black hole mass approaching @xmath110 , as the missing core light suggests ( see section 4.6 ) , could power the outburst by bondi - like accretion for @xmath111 @xmath99 and a central gas temperature approaching @xmath112 kev within one arcsec of the nucleus .
these conditions would be accompanied by a four - fold increase in surface brightness in the inner arcsec of the x - ray image .
no evidence of this is seen , although a suitable combination variables may be found that could match the observation .
the key here is whether indeed an ultramassive black hole lies in the nucleus of ms0735 .
its unusually large core hints at this remarkable possibility . but direct evidence for smbhs substantially in excess of @xmath113 remains elusive .
the efficiency of stellar accretion onto smbhs is governed by the two body relaxation timescale , @xmath114 , which is much longer than the age of the universe in the low density cores of bcgs .
a feedback scenario involving the capture of stars by a nuclear smbh has been discussed by wang & hu ( 2005 ) . using the methodology of syer & ulmer ( 1999 ) ,
wang & hu found capture rates in bulges with @xmath47 to @xmath115 black holes to be @xmath116 to @xmath117 , giving agn energies of @xmath118 per outburst .
these energies lie far below the output of ms0735 .
miralda - escud & kollmeier ( 2005 ) have argued that stellar capture can be enhanced by the presence of an accretion disk . but this model still requires a sufficiently dense reservoir of stars to capture , which appears to be absent in ms0735 s large , low surface brightness core .
stellar accretion seems to be an unlikely source of accreted mass .
cooling flows often harbor greater than @xmath119 of molecular gas ( edge 2001 , salome & combes 2003 ) , providing a reservoir from which to fuel star formation and the agn ( pizzolato & soker 2005 ) .
given the large nuclear accretion rate implied by the agn power , we would expect the nucleus to be awash in gas and star formation , which is clearly not the case . while the upper limits on the central gas mass presented in section 3.3 can not exclude the existence of gas in a less visible form ,
the observed limit of @xmath120 is uncomfortably close to the @xmath121 of gas required to fuel the outburst by accretion .
accretion onto an agn is an inefficient process that is usually accompanied by star formation .
the existence of the magorrian relation shows that the accretion efficiency onto smbhs historically is roughly one part in seven hundred .
some relief can be found when the mass to energy conversion efficiency approaches @xmath122 for a rapidly - spinning smbh .
but it would not fix the problem .
it is just as likely , if not more so , that @xmath123 .
allen et al . ( 2006 ) and merloni & heinz ( 2007 ) found @xmath48 to be closer to a few percent , rather than the canonical 10% we have assumed .
natarajan & treister ( 2008 ) found an average efficiency of roughly 5% , based on the shape of the x - ray luminosity function .
regardless of the assumed efficiency , fueling the agn by cold accretion demands an unrealistically large fraction of the cold gas supply at the center the bcg to be channeled onto the black hole in only @xmath47 yr .
two gas - rich systems from edge s sample we have studied in detail are abell 1068 and abell 1835 ( mcnamara et al .
. both harbor star formation at rates approaching or exceeding @xmath124 and contain @xmath125 of molecular gas .
the agn power in abell 1068 is quite modest , and most of its cold , molecular gas is being consumed by star formation .
only a small fraction of the gas in these systems can be accreting onto the smbh .
the central galaxy in the abell 1835 cluster is cloaked in @xmath126 of molecular gas , and star formation is proceeding at a rate of @xmath127 . yet
its agn power , and hence the accretion rate onto its smbh , is about an order of magnitude smaller than ms0735 s . in this system , for every unit of mass accreting onto the smbh
, 300600 units are consumed by star formation , a ratio that is close to the expectation of the magorrian relation . in ms0735
the situation is reversed . between 3 and 20 units of mass
are accreting onto the smbh for every one consumed by stars .
this requires a channeling mechanism @xmath128 to @xmath88 times more efficient than the mechanisms operating in gas - rich cooling flows and during the early stages of galaxy formation .
assuming for the moment that cold gas is accreting onto the smbh at its nominal rate in a spherical flow proceeding essentially at the free - fall speed , @xmath129 , the radius at which the surface density of gas will exceed the critical surface density for the onset of molecular cloud condensation and star formation @xmath130 ( martin & kennicutt 2001 ) is @xmath131 here , @xmath129 is set to the typical one dimensional velocity dispersion of a bcg , and @xmath132 is the accretion rate onto the smbh .
this value is comparable to the gravitational radius @xmath133 pc for a @xmath43 black hole .
so it is possible that the gas has fallen into the black hole before stars were able to form .
however , the assumptions going into this calculation are exceedingly , and perhaps unrealistically , optimistic .
they are at variance with what is observed in otherwise similar systems .
it is difficult to imagine all of the gas in the halo suddenly flowing radially inward near to the free - fall speed without forming a disk and stars .
but we have not ruled it out .
ms0735 is among the few objects whose jet energy and power levels are so demanding that the spin and accretion mechanisms can be meaningfully constrained .
accretion and spin are distinct in their parametric dependencies on power and energy .
for spin , the _ energy _ output depends primarily on black hole mass and spin parameter , while for accretion , the output _ energy _ depends primarily on the total accreted mass .
spin _ power _
depends primarily on the squares of the poloidal magnetic field strength ( pressure ) and spin parameter , while accretion _
power _ depends primarily on mass accretion rate .
these dependencies allow us , in principle , to distinguish between the two mechanisms .
we showed earlier that accretion seems unable to power this system , and we suggest a spinning smbh as a plausible alternative power source .
a rapidly rotating smbh with a reasonable mass contains enough energy and , by applying a sufficient torque , enough power to launch and maintain the jet in ms0735 .
here we adopt the energy and power scaling relations of meier s ( 1999 , 2001 ) hybrid model to compare to our energy and power measurements .
@xmath134 and @xmath135 where @xmath136 is the spin parameter that varies between 0 for a non spinning black hole and 1 for a maximally spinning hole , @xmath137 is the poloidal magnetic field strength threading the the accretion disk and ergosphere , and @xmath138 is the black hole mass in units of @xmath7 . in this formalism , spin energy is transformed into jet power through a torque applied by @xmath139 .
the estimated jet power is related to the accretion rate through the requirement that the disk can support the associated magnetic stress ( eg . , meier 2002 ) .
assuming @xmath140 and adopting a smbh of mass @xmath43 , meier s spin model yields an energy of @xmath141 erg . this value is larger than the @xmath142 found from the cavities and shock fronts . achieving a jet power of @xmath143 with these parameters
requires a magnetic field strength near the hole of @xmath144 g. this appears to be a plausible value within the context of meier s theory .
the formulas above allow a great deal of latitude in @xmath137 given the poor constraints on the mass and spin of the hole , and there is a good deal of uncertainty in the theory itself .
provided a means of tapping its power is operational , black hole spin provides a plausible alternative to pure accretion power .
it is noteworthy that the spin mechanism requires some accretion in order to maintain the magnetic field that couples spin to jet power .
because the liberated energy is transformed primarily from the rotation of the hole itself , the spin model requires an accretion rate that can be a small fraction @xmath145 of the rate required to power a jet by pure accretion ( meier 1999 ) .
in other words , unless the energetic output from spin substantially exceeds the accretion power , our argument for spin powering is largely moot . whether or not the blandford - znajek mechanism and its variants
are able to accomplish this is unclear ( eg . , ghosh & abramowicz 1997 ) . however , turning this point around , the conditions in ms0735 strongly suggest that the energetic output from spin exceeds the output from accretion by a large factor
. this would have important consequences for ms0735 in particular and giant ellipticals centered in cooling atmospheres in general .
if accretion from cooling atmospheres surrounding smbhs provides the gas required to tap black hole spin power , agn would be effectively coupled to the atmospheres by a feedback loop .
agn feedback may be the mechanism that prevents large cooling flows from forming ( mcnamara & nulsen 2007 ) . the spin model is not without its problems .
we do not understand how the smbh spun - up , and if it was born spinning , why its spin was tapped only recently .
a smbh can be spun - up by the accretion of high angular momentum gas , or the accretion of one or more smbhs ( wilson & colbert 1995 , volonteri et al .
both mechanisms require an accreted mass that is comparable to the progenitor s mass to achieve a spin parameter much greater than @xmath56 ( rezzolla et al .
one problem with this scenario as it applies to ms0735 is that typical galaxy mergers at late times occur with a roughly 10:1 mass ratio ( sesana et al .
2004 ) , implying a similar ratio between their smbhs .
ms0735 s bcg is among the most massive galaxies in the universe
. it would be unlikely to find a galaxy of comparable mass to merge with at late times .
buildup through mergers of smaller units during the hierarchical growth of galaxies is not expected to yield high spin factors ( hughes & blandford 2003 , berti & volonteri 2008 ) .
gas accretion appears to be most efficient at spinning - up a hole ( moderski & sikora 1996 ) .
a potential problem with this scenario is the absence of a stellar cusp , which indicates that the bcg has not accreted an appreciable amount of gas at late times . it may be possible to estimate the the size of the black hole or holes that merged to create ms0735 s smbh through the size of its core .
cusp - like surface brightness profiles are generally found in elliptical galaxies fainter than @xmath146 ( cot et al .
2006 , balcells et al .
2007 ) , while luminous ellipticals like ms0735 generally have cores ( kormendy 1985 , laine et al .
2003 , lauer et al .
this dichotomy may be driven by gas and stellar dynamics during galaxy formation and evolution . as gas cools and sinks to the centers of dark matter halos , dissipation channels it to the nucleus creating a stellar cusp , not a core .
cusps may form by primordial cooling and subsequently during gas - rich mergers ( faber et al .
1997 , kormendy et al .
2008 ) . as lower luminosity galaxies merge into larger entities via dissipationless `` dry '' mergers ,
their cusps are expected to be preserved ( milosavljevic & merritt 2001 ) .
cores , or light deficits , are thought to be created by dynamical post - processing through interactions such as scouring by accreting smbhs ( faber et al .
1997 , graham 2004 , gualandris & merritt 2007 , lauer et al .
2007 , kormendy et al . 2008 ) , or perhaps by baryon oscillations driven by agn outbursts ( peirani et al . 2008 ) .
black hole scouring operates by transferring energy from the inspiraling black hole to the stars in the light cusp through dynamical friction ( milosavljevic & merritt 2001 , milosavljevic et al .
missing light in the nucleus , converted to appropriate mass units , is expected to be on the order of the sum of the masses of the inspiraling and host smbhs .
gualandris & merritt ( 2007 ) estimated the light deficit by extrapolating a sersic profile inward to the nucleus from the break radius .
we have done this for ms0735 , and the results are shown in fig .
9 . a sersic profile ( solid line )
is fited to the light profile between the innermost extreme value of the break radius and the end of usable data at about @xmath147 arcsec . beyond the break radius
the sersic profile closely follows the @xmath148-law profile .
taking the difference between the observed profile and the extrapolated sersic law , we find a `` missing light '' of @xmath149 .
adopting an @xmath23-band mass - to - luminosity ratio of 3 , which is a typical value for giant elliptical galaxies , we find a very rough estimate of the mass deficit to be @xmath110 .
this figure is a few times larger than the largest deficits found by lauer et al .
( 2007a ) , and is much larger than those considered by kormendy & bender ( 2009 ) , reflecting the unusually large size of ms0735 s break radius .
the unknown shape of the progenitor profile is a large source of systematic uncertainty .
gulandris & merritt ( 2007 ) have argued that the light deficit method breaks down when the core radius exceeds the gravitational radius of influence of the central black hole .
the expression for the radius of influence , @xmath150 , where @xmath151 is the smbh mass expressed in units of @xmath152 ( ferrarese & ford 2005 ) , yields a value of @xmath153 pc for a black hole mass of @xmath43 .
this figure is much smaller than its 3.1 kpc break radius .
therefore , the scouring model may not offer a complete explanation .
kormendy & bender ( 2009 ) have shown that mass deficit correlates with smbh mass in several nearby systems , and lauer et al .
( 2007a , b ) proposed that oversize cores witness the presence of `` ultramassive '' black holes exceeding @xmath119 .
these studies certainly add grist to our conjecture that ms0735 harbors such an ultramassive black hole . however , caution is in order .
ms0735 is uncharted territory .
how such a large core and enormous black hole would be created is a mystery .
one possibility is slingshot ejection by a massive binary or a merger kick ( eg . , boylan - kolchin et al . 2004 , bogdanovi et al .
if some sort of scouring took place , our mass deficit measurement indicates that an ultramassive black hole would have been involved .
the possible existence of ultramassive black holes and their numbers are of great interest as they provide an upper limit to the amount of energy deposited in hot atmospheres over the ages of galaxies and clusters ( fujita & reiprich 2004 , mcnamara & nulsen 2007 ) , and their numbers would constrain the mass function of seed black holes at high redshift ( natarajan & treister 2008 ) . drawing on an analogy to
the `` low - hard '' , `` high - soft '' states seen in black hole binaries , gallo et al .
( 2003 ) , merloni et al .
( 2003 ) , falke et al .
( 2004 ) , churazov et al .
( 2005 ) and others have argued that mechanical outflows and jets operate at low accretion rates @xmath154 . in black hole x - ray binaries , this corresponds to the `` low hard '' state . at high accretion rates
@xmath155 , the `` high - soft '' regime , a geometrically thin , optically thick disk forms , and most of the accretion energy is released in the form of radiation and winds emanating from the accretion disk , rather than in a jet .
ms0735 s implied accretion rate @xmath156 , places it near to the high - soft regime . in the unified scenario , its nucleus should be as bright as a quasar .
it is not .
neither the adaf or adios models should be operational at such a high accretion rate , so any radiation generated in a disk should be evident . the size of this discrepancy is shown in fig . 10 , where we have adapted the cartoon sketch of the relationship between accretion rate and agn power given by churazov et al .
this figure shows the expected partitioning between radiation from the accretion disk and mechanical jet power as a function of mass accretion rate . in the low - hard state where @xmath157 , agn power is dominated by a jet and disk radiation is negligible . in the high - soft state ,
@xmath158 the agn power is dominated by radiation from the disk .
our measurements of ms0735 s mechanical ( heating ) power and the upper limit to the disk radiation are indicated on the plot .
ms0735 s energy output is dominated by mechanical / heating energy and not radiation , which is inconsistent with the unified model .
if the black hole is indeed ultramassive and greatly exceeds @xmath119 , as its core size suggests , ms0735 may fall into the low - hard , regime thus bringing it into consistency with the model .
how massive the hole must be to accomplish this is unclear .
the location of the critical accretion rate where the agn output power transitions from mechanical output to radiation is poorly understood ( eg .
, wu & cao 2008 ) , and may be as low as @xmath159 ( fender et al .
a lower transition value would strengthen the case that ms0735 does not fit the accretion paradigm .
finally , one can imagine we are seeing a post accretion event where the quasar recently shut off .
but this seems both fortuitous and implausible as it would not change the incredibly efficient demands on accretion or the lack or star formation . unless ms0735 s black hole is substantially in excess of the magorrian value , it nuclear emission properties are inconsistent with with the accretion paradigm
we have shown that using measurements of x - ray cavities and shock fronts , gas masses , and structural parameters of the host galaxy , we are able to levy interesting constraints on smbh mass and spin . unless a large and unseen reservoir of gas is flowing onto the nucleus with a remarkably high efficiency , and unless its black hole mass dramatically exceeds @xmath109 , it would be difficult to power ms0735 s agn by accretion alone . adding to this , the bcg s nuclear luminosities at optical , uv , and x - ray wavelengths lie several orders of magnitude below the expected accretion luminosities at the rate required to fuel its agn . in light of these problems , we have shown that its agn may be powered by a spinning , @xmath160 to @xmath109 black hole for a reasonable value of its spin parameter .
our argument hinges on the assumption that the spin power released greatly exceed the accretion power of the gas maintaining the magnetic field pressure near the black hole .
ms0735 s extreme properties have interesting consequences for feedback models operating in clusters and galaxies at late times ( eg . , croton et al .
2006 , pizzolato & soker 2005 ) . in order to operate a spin - powered feedback loop ,
its energy must be tapped continuously , or nearly so , if it is to maintain balance between heating and cooling .
accretion at some level is needed to confine the poloidal magnetic field that torques the spinning black hole and taps its power ( cf .
, nemmen et al . 2007 ) .
if a relatively modest level of hot or cold accretion from the surrounding x - ray atmosphere is responsible , it could provide the link between the cooling atmosphere and agn power that maintains a self - regulating feedback loop ( cf .
mcnamara & nulsen 2007 ) .
ms0735 raises the possibility that both spin and accretion are important feedback modes in clusters .
spin - regulated feedback , or accretion power boosted by spin , may explain objects such as awm 4 ( osullivan et al .
2005 , giacintucci et al .
2008 , gastaldello et al . 2008 ) and abell 2029 ( clarke et al . 2004 ) whose nuclei are devoid of large reservoirs of cold gas and star formation , and yet apparently are able to maintain sufficient agn power output to stave - off a cooling flow . finally , the need to invoke spin power may be avoided if the black hole mass dramatically exceeds @xmath109 , a remarkable possibility that finds support in the bgc s unusually large core radius .
this notion is highly speculative , but worth considering given the tight correlation between central light deficit and smbh mass found by kormendy & bender ( 2009 ) in lower luminosity systems .
the radius of gravitational influence of a @xmath110 black hole is @xmath161 pc , which corresponds to an angular size of @xmath162 arcsec at the distance of ms0735 .
therefore , it would be possible to detect or place interesting limits on the existence of such a massive black hole .
how such an object would form and the fate of the enormous binding energy it would release is an open and interesting question .
we thank tod lauer and the referee mateusz ruszkowski for comments that improved the paper , and we acknowledge insightful discussions with dan evans , bob oconnell , megan donahue , genevieve de messires and hui li .
we thank eugene churazov and bill forman for permission to reproduce their diagram shown in fig .
10 , and mina rohanizadegan for assistance with the analysis .
this work was supported by generous grants from nasa and the natural sciences and engineering research council of canada .
bower , r. g. , benson , a. j. , malbon , r. , helly , j. c. , frenk , c. s. , baugh , c. m. , cole , s. , & lacey , c. g. 2006 , , 370 , 645 boylan - kolchin , m. , ma , c .- p . , & quataert , e. 2004 , , 613 , l37 brggen , m. , & kaiser , c. r. 2001 , , 325 , 676 croston , j. h. , hardcastle , m. j. , birkinshaw , m. , worrall , d. m. , & laing , r. a. 2008 , , 386 , 1709 croton , d. j. , et al .
2006 , , 365 , 11 de young , d. s. 2006 , , 648 , 200 de messires , g. , oconnell , r. w. , mcnamara , b. r. , donahue , m. , nulsen , p. e. j. , voit , m. , & wise , m. w. 2009 , apj , submitted ferrarese , l. , & merritt , d. 2000 , , 539 , l9 fujita , y. , & reiprich , t. h. 2004 , , 612 , 797 gallo , e. , fender , r. p. , & pooley , g. g. 2003 , , 344 , 60 gastaldello , f. , buote , d. a. , brighenti , f. , & mathews , w. g. 2008 , , 673 , l17 jones , t. w. , & de young , d. s. 2005 , , 624 , 586 kormendy , j. 1985 , , 292 , l9 kormendy , j. et al .
2008 , apj , submitted .
kormendy , j. , & bender , r. 2009 , , 691 , l142 laine , s. , van der marel , r. p. , lauer , t. r. , postman , m. , odea , c. p. , & owen , f. n. 2003 , , 125 , 478 lauer , t. r. , et al . 2007b , , 664 , 226 lauer , t. r. , et al .
2007a , , 662 , 808 li , h. , lapenta , g. , finn , j. m. , li , s. , & colgate , s. a. 2006 , , 643 , 92 martin , c. l. , & kennicutt , r. c. , jr .
2001 , , 555 , 301 mathews , w. g. , & brighenti , f. 2008 , , 685 , 128 marconi , a. , & hunt , l. k. 2003 , , 589 , l21 mccarthy , i. g. , balogh , m. l. , babul , a. , poole , g. b. , & horner , d. j. 2004 , , 613 , 811 mcnamara , b. r. , & nulsen , p. e. j. 2007 , , 45 , 117 mcnamara , b. r. , et al .
2000 , , 534 , l135 mcnamara , b. r. , nulsen , p. e. j. , wise , m. w. , rafferty , d. a. , carilli , c. , sarazin , c. l. , & blanton , e. l. 2005 , , 433 , 45 mcnamara , b. r. , et al .
2006 , , 648 , 164 meier , d. l. 1999 , , 522 , 753 meier , d. l. 2001 , , 548 , l9 meier , d. l. 2002 , new astronomy review , 46 , 247 miralda - escud , j. , & kollmeier , j. a. 2005 , , 619 , 30 merloni , a. , heinz , s. & di matteo , t. , 2003 , mnras , 345 , 1057 merloni , a. , & heinz , s. 2007 , , 381 , 589 milosavljevi , m. , & merritt , d. 2001 , , 563 , 34 milosavljevi , m. , merritt , d. , rest , a. , & van den bosch , f. c. 2002 , , 331 , l51 moderski , r. , & sikora , m. 1996 , , 283 , 854 nakamura , m. , li , h. , & li , s. 2007 , , 656 , 721 nakamura , m. , tregillis , i. l. , li , h. , & li , s. 2008 , arxiv e - prints , 806 , arxiv:0806.4150 natarajan , p. , & treister , e. 2008 , arxiv:0808.2813 nemmen , r. s. , bower , r. g. , babul , a. , & storchi - bergmann , t. 2007 , , 377 , 1652 nulsen , p. e. j. , & fabian , a. c. 2000 , , 311 , 346 nulsen , p. e. j. , jones , c. , forman , w. r. , david , l. p. , mcnamara , b. r. , rafferty , d. a. , brzan , l. , & wise , m. w. 2007 , heating versus cooling in galaxies and clusters of galaxies , 210 nusser , a. , silk , j. , & babul , a. 2006 , , 373 , 739 peterson , j. r. , & fabian , a. c. 2006 , , 427 , 1 osullivan , e. , vrtilek , j. m. , kempner , j. c. , david , l. p. , & houck , j. c. 2005 , , 357 , 1134 rezzolla , l. , barausse , e. , dorband , e. n. , pollney , d. , reisswig , c. , seiler , j. , & husa , s. 2008 , , 78 , 044002 risaliti , g. , maiolino , r. , & salvati , m. 1999 , , 522 , 157 ruszkowski , m. , brggen , m. , & begelman , m. c. 2004 , , 611 , 158 salim , s. , et al . 2007 , , 173 , 267 salom , p. , & combes , f. 2003 , , 412 , 657 salom , p. , & combes , f. 2008 , arxiv e - prints , 806 , arxiv:0806.4545 sesana , a. , haardt , f. , madau , p. , & volonteri , m. 2004 , , 611 , 623 lcccc a & 1.51 & [email protected] & [email protected] & [email protected] + b & 1.74 & [email protected] & [email protected] & [email protected] + c & 0.988 & [email protected] & [email protected] & [email protected] + d & 0.581 & [email protected] & [email protected] & [email protected] + e & 0.0872 & [email protected] & [email protected] & [email protected] + f & 0.494 & [email protected] & [email protected] & [email protected] + g & 1.74 & [email protected] & [email protected] & [email protected] + h & 0.0901 & [email protected] & [email protected] & [email protected] + i & 5.64 & [email protected] & [email protected] & [email protected] + | powering the @xmath0 nuclear outburst in the ms0735.6 + 7421 cluster central galaxy by accretion with a 10% mass - to - energy conversion efficiency implies that its putative supermassive black hole ( smbh ) grew by @xmath1 over the past 100 myr .
guided by data at several wavelengths , we place upper limits on the amount of cold gas and star formation near the nucleus of @xmath2 and @xmath3 , respectively .
these limits imply that an implausibly large fraction of the preexisting cold gas in the inner several kpc must have been consumed by its smbh at the rate of @xmath4 during the past 100 myr while leaving no trace of star formation .
such a high accretion rate would be difficult to maintain by stellar accretion or the bondi mechanism , unless the black hole mass approaches @xmath5 .
furthermore , its feeble nuclear luminosities in the uv , i , and x - ray bands compared to its enormous mechanical power are inconsistent with rapid accretion onto a @xmath6 black hole .
we suggest instead that the agn outburst is powered by angular momentum released from a rapidly - spinning black hole .
the rotational energy and power available from a spinning black hole are consistent with the cavity and shock energetics inferred from x - ray observations .
a maximally - spinning , @xmath7 black hole contains enough rotational energy , @xmath8 erg , to quench a cooling flow over its lifetime and to contribute significantly to the excess entropy found in the hot atmospheres of groups and clusters .
two modes of agn feedback may be quenching star formation in elliptical galaxies centered in cooling halos at late times .
an accretion mode that operates in gas - rich systems , and a spin mode operating at modest accretion rates .
the spin conjecture may be avoided in ms0735 by appealing to bondi accretion onto a central black hole whose mass greatly exceeds @xmath9 .
the host galaxy s unusually large , 3.8 kpc stellar core radius ( light deficit ) may witness the presence of an ultramassive black hole .
= 1 |
in this paper , every graph is finite and simple . for a graph @xmath6
, a bijection @xmath7 is an _ automorphism _ of @xmath6 if @xmath8 if and only if @xmath9 .
for example , the identity function on @xmath10 is an automorphism of @xmath6 .
we call this identity function the _ trivial _ automorphism . for a tree
@xmath0 with @xmath1 , we say @xmath0 is _ automorphism - free _ if there is no non - trivial automorphism of @xmath0 . for instance , the following graph @xmath11 in figure 1 is automorphism - free .
it is easy to check that there are no automorphism - free trees with fewer than 7 vertices .
let @xmath2 be the poset ( partially ordered set ) with the element set of all finite automorphism - free trees ( up to graph isomorphism ) ordered by @xmath3 if @xmath4 can be obtained from @xmath5 by successively deleting one leaf at a time in such a way that each intermediate tree is also automorphism - free . in this paper , we prove that @xmath2 has a unique minimal element , namely @xmath11 .
[ main1 ] let @xmath0 be a minimal element of the poset @xmath2 .
then @xmath0 is isomorphic to @xmath11 .
equivalently , [ main2 ] every automorphism - free tree @xmath0 can be obtained from @xmath11 by successively adjoining a leaf at a time in such a way that each intermediate tree is also automorphism - free . or , [ main3 ] for every automorphism - free tree @xmath0 , @xmath11 can be obtained from @xmath0 by successively deleting a leaf at a time in such a way that each intermediate tree is also automorphism - free .
this result gives an affirmative answer to the question asked by rupinski in @xcite .
first , we start with some definitions .
a _ component _ of a graph @xmath6 is a maximal non - null subgraph of @xmath6 . for a vertex @xmath12 of a graph @xmath6
, @xmath13 denotes the graph obtained from @xmath6 by deleting the vertex @xmath12 ( deleting all the edges incident with @xmath12 as well ) . for an edge @xmath14 of a graph @xmath6
, @xmath15 denotes the graph obtained from @xmath6 by deleting the edge @xmath14 ( not deleting the vertex @xmath12 or @xmath16 ) . for a vertex set @xmath17 of a graph @xmath6
, @xmath18 denotes the subgraph of @xmath6 induced by @xmath19 . for a tree @xmath0 , a _ leaf _
@xmath20 is a vertex of degree one in @xmath0 , and @xmath21 denotes the ( unique ) neighbor of @xmath20 in @xmath0 . for a path @xmath22 ,
the _ length _ of @xmath22 is the number of edges in @xmath22 . for a tree @xmath0 and @xmath23 ,
@xmath24 is the length of the ( unique ) path from @xmath12 to @xmath16 in @xmath0 . for each @xmath25 ,
we say @xmath25 is a _ center _ of @xmath0 if @xmath27 for every @xmath28 , and the _ radius _ @xmath29 of @xmath0 denotes the number @xmath30 . for the proof of [ main1 ]
, we look at a minimal element @xmath0 of @xmath2 . in @xmath0
, we choose special leaves @xmath31 and @xmath32 by certain methods , and use the fact that both @xmath33 and @xmath34 have non - trivial automorphisms . from this , we find various properties that @xmath0 must have . for instance , we prove that @xmath0 must have two centers , and @xmath33 must have exactly one center , and @xmath34 must have two centers , etc . eventually we prove that @xmath0 must be isomorphic to @xmath11 .
the following is an easy lemma about centers in a tree .
we omit the proof . * if @xmath12 and @xmath16 are distinct centers of @xmath0 , then @xmath36 .
in particular , there are at most two centers of @xmath0 . *
if @xmath12 and @xmath16 are distinct centers of @xmath0 , then every path of length @xmath29 from @xmath12 contains @xmath16 , and vice versa . * if @xmath12 is the unique center of @xmath0 , then @xmath37 . *
if @xmath12 and @xmath16 are distinct centers of @xmath0 , then @xmath35 either fixes @xmath12 and @xmath16 or switches them . *
if @xmath12 is the unique center of @xmath0 , then it is a center of @xmath38 as well . *
if @xmath12 and @xmath16 are distinct centers of @xmath0 , then every center of @xmath38 is either @xmath12 or @xmath16 .
let @xmath22 be the path from @xmath12 to @xmath20 in @xmath0 and number the vertices of @xmath22 as @xmath40 in order . for each @xmath41 ,
let @xmath42 be the component of @xmath43 containing @xmath20 .
then for every component @xmath44 of @xmath43 not containing @xmath12 , @xmath45 . * proof .
* we proceed by induction on @xmath46 .
it is easy to see that the statement holds for @xmath47 .
consider all neighbors of @xmath12 .
each one is in its own component of @xmath48 . among those components ,
we take one with the least number of vertices .
( if there is more than one smallest component , just pick any one of them . )
let @xmath44 be the component of @xmath48 we chose and let @xmath16 be the neighbor of @xmath12 in @xmath44 .
now , look at all children of @xmath16 ( the neighbors of @xmath16 in @xmath44 ) .
if there are no children of @xmath16 , then @xmath16 is a special leaf with respect to @xmath0 and @xmath12 we are looking for .
therefore we may assume @xmath49 .
then from the induction hypothesis , there is a special leaf @xmath20 with respect to @xmath44 and @xmath16 .
then , it is easy to check that @xmath20 is a special leaf with respect to @xmath0 and @xmath12 as well .
this proves [ special ] .
* for the sake of contradiction , suppose @xmath0 has only one center @xmath12 .
let @xmath31 be the special leaf with respect to @xmath0 and @xmath12 .
let @xmath50 and take a non - trivial automorphism @xmath35 of @xmath51 . by [ centers ] ( 5 ) , @xmath12 is a center of @xmath51 as well .
\(1 ) _ @xmath35 does not fix @xmath12 .
in particular , there is another center of @xmath51 .
_ suppose @xmath35 fixes @xmath12 .
notice that @xmath35 does not fix @xmath52 because otherwise we can extend @xmath35 to a non - trivial automorphism of @xmath0 by assigning @xmath53 .
in particular , @xmath54 .
let @xmath22 be the path from @xmath12 to @xmath52 in @xmath51 , and number the vertices of @xmath22 as @xmath55 in order ( @xmath56 ) .
let @xmath57 be the largest number such that @xmath58 ( and so , @xmath59 and @xmath60 ) . in @xmath61 ,
let @xmath62 be the component containing @xmath63 ( this component contains @xmath52 as well ) , and let @xmath64 be the component containing @xmath65 .
clearly , @xmath62 and @xmath64 are different since @xmath63 and @xmath65 are both neighbors of @xmath66 .
notice that @xmath67 is an automorphism of @xmath61 since @xmath35 fixes @xmath66 .
this implies @xmath62 and @xmath64 are isomorphic
. in particular , @xmath68 . but back in @xmath0 , @xmath69 and @xmath64 are two components of @xmath70 , and @xmath71 .
this contradicts the definition of @xmath31 .
this proves ( 1 ) .
@xmath72 , and there is a unique path of length @xmath30 from @xmath16 in @xmath0 , namely the path from @xmath16 to @xmath31 . _ in @xmath0 , @xmath12 is the unique center .
but in @xmath51 , both @xmath12 and @xmath16 are centers
. therefore @xmath73 in particular , @xmath74 .
this implies there is no path of length @xmath30 from @xmath16 in @xmath51 .
since there is a path of length @xmath30 from @xmath16 in @xmath0 , namely the path from @xmath16 to @xmath31 , it must be unique .
this proves ( 2 ) .
\(3 ) _ @xmath31 is not a neighbor of @xmath12 .
_ suppose @xmath31 is a neighbor of @xmath12 . then from ( 2 ) , @xmath75 since @xmath51 has two centers and @xmath76 , @xmath77 and @xmath78 . but
this is impossible since there is no automorphism - free tree with three vertices .
this proves ( 3 ) .
@xmath52 has degree two in @xmath0 .
_ if there exists another child @xmath79 of @xmath52 in @xmath0 , then the path from @xmath16 to @xmath79 is another path of length @xmath30 from @xmath16 in @xmath0 , which is impossible by ( 2 ) .
therefore @xmath31 is the unique child of @xmath52 in @xmath0 .
this proves ( 4 ) .
let @xmath80 and @xmath81 be the two components of @xmath82 containing @xmath12 and @xmath16 , respectively . note that @xmath35 switches @xmath12 and @xmath16 by ( 1 ) . and @xmath83 is isomorphic to @xmath81 by @xmath35 . let @xmath32 be a special leaf with respect to @xmath81 and @xmath16 ( @xmath32 exists since @xmath84 ) .
let @xmath85 and take a non - trivial automorphism @xmath86 of @xmath87 .
\(5 ) _ @xmath12 is a center of @xmath87 , and @xmath16 is not . _ again , @xmath12 is a center of @xmath87 by [ centers ] ( 5 ) .
but , @xmath16 is not a center of @xmath87 because from ( 2 ) , @xmath88 this proves ( 5 ) .
\(6 ) _ @xmath86 does not fix @xmath16 .
moreover , @xmath89 .
_ suppose @xmath86 fixes @xmath16 .
then @xmath12 is fixed as well because among the neighbors of @xmath16 , @xmath12 is the unique center of @xmath87 ( although @xmath12 might not be the unique center of @xmath87 ) . since both @xmath12 and @xmath16
are fixed by @xmath86 , @xmath90 .
therefore @xmath91 is a non - trivial automorphism of @xmath92 ( otherwise we can extend @xmath86 to a non - trivial automorphism of @xmath0 by assigning @xmath93 ) .
then by the same argument as in ( 1 ) , this contradicts the definition of @xmath32 .
since @xmath16 is adjacent to a center of @xmath87 , so is @xmath94 .
and @xmath16 is the unique such vertex in @xmath95 .
therefore @xmath94 does not belong to @xmath95 because @xmath96 .
also , every member of @xmath97 does not belong to @xmath95 either because @xmath92 is a component of @xmath98 containing @xmath16 .
therefore @xmath89 .
this proves ( 6 ) .
\(8 ) _ @xmath109 . in particular
, the degree of @xmath12 is two ( both in @xmath87 and @xmath0 ) , and @xmath110 consists of two components @xmath111 and @xmath92 . _ from ( 3 )
, @xmath12 has no neighbor which is a leaf ( both in @xmath0 and @xmath87 ) .
in particular , every component of @xmath98 has more than one vertex .
note that the union of all components of @xmath110 different from @xmath112 has size @xmath113 .
since @xmath44 is one of them whose size is at least @xmath102 , there can not be another one ( and so , @xmath114 ) .
therefore @xmath12 has degree two , and this proves ( 8) .
\(9 ) _ @xmath86 does not fix @xmath12 .
in particular , @xmath87 has two centers @xmath12 and @xmath116 , and @xmath117 .
_ by ( 5 ) , @xmath16 is not a center of @xmath87 and it is adjacent to a center of @xmath87 .
therefore @xmath94 has the same property in @xmath87 .
but @xmath94 is not adjacent to the center @xmath12 , because if it is , then @xmath118 is isomorphic to @xmath92 and hence , @xmath119 .
this is impossible by ( 8) .
therefore @xmath94 is adjacent to another center , and this also implies @xmath120 . since two centers are adjacent , @xmath116 must be the neighbor of @xmath12 different from @xmath16 .
since @xmath116 also has degree two , the neighbor of @xmath116 different from @xmath12 is @xmath94 . and
@xmath112 is isomorphic to @xmath121 .
this proves ( 9 ) .
\(10 ) _ @xmath80 and @xmath81 are paths .
_ it is enough to show that @xmath80 is a path since @xmath124 .
suppose there exists a vertex of degree at least three in @xmath80 .
choose such a vertex @xmath125 with @xmath126 as small as possible .
then , @xmath127 has degree at least three in @xmath80 as well . to see this , first observe that @xmath128 has degree at least three since @xmath129 , and @xmath123
. therefore @xmath130 has degree at least three since @xmath124 .
but then , @xmath131 this contradicts our choice of @xmath79 .
this proves ( 10 ) .
since both @xmath12 and @xmath16 have degree two in @xmath0 , @xmath0 is a path by ( 10 ) .
this contradicts the fact that @xmath0 is automorphism - free .
therefore @xmath0 has two centers .
this proves [ aft1 ] . *
proof of [ main1 ] .
* by [ aft1 ] , @xmath0 has two centers @xmath12 and @xmath16 .
let @xmath80 and @xmath81 be the two components of @xmath82 containing @xmath12 and @xmath16 , respectively .
let @xmath31 be a special leaf with respect to @xmath80 and @xmath12 and let @xmath32 be a special leaf with respect to @xmath81 and @xmath16 .
let @xmath132 be the shortest distance from @xmath12 to a vertex in @xmath80 whose degree in @xmath0 is at least three . if there is no such vertex , then set @xmath132 as @xmath133 .
similarly , let @xmath134 be the shortest distance from @xmath16 to a vertex in @xmath81 whose degree in @xmath0 is at least three .
without loss of generality , we may assume @xmath135 . and
further we may assume if @xmath136 then @xmath137 by switching @xmath12 and @xmath16 if necessary .
we first consider @xmath50 .
let @xmath35 be a non - trivial automorphism of @xmath51 .
@xmath12 and @xmath16 are centers of @xmath51 . _ by [ centers ] ( 2 ) , @xmath138 since every path of length @xmath139 from @xmath12 in @xmath0 does not contain @xmath31 . and by [ centers ] ( 6 ) , @xmath16 is a center of @xmath51 since @xmath140 .
again by [ centers ] ( 6 ) , if there is a center of @xmath51 different from @xmath16 , then it must be @xmath12 .
for the sake of contradiction , suppose @xmath16 is the unique center of @xmath51 .
then @xmath52 is a leaf in @xmath51 by the same argument as in ( 4 ) in the proof of [ aft1 ] .
then , @xmath35 does not fix @xmath12 since otherwise it contradicts the definition of @xmath31 , by the same argument as in ( 6 ) in the proof of [ aft1 ] .
therefore in @xmath141 , the component @xmath83 is isomorphic to another component @xmath44 of @xmath141 .
note that @xmath142 since @xmath106 is a subset of @xmath143 , @xmath144 together with our assumption @xmath135 , @xmath145 .
moreover @xmath141 has exactly two components , namely @xmath44 and @xmath83 .
in particular , @xmath16 has degree two in @xmath0 .
next , there exists a vertex in @xmath146 whose degree in @xmath0 is at least three because otherwise @xmath81 is a path , and hence so is @xmath51 , and so is @xmath0 because @xmath52 has degree two in @xmath0 .
but then , @xmath147 by @xmath35 , and this contradicts our assumption that @xmath137 if @xmath136 .
therefore @xmath16 is not the unique center of @xmath51 .
this implies @xmath12 is a center of @xmath51 as well .
this proves ( 1 ) .
\(2 ) _ @xmath35 switches @xmath12 and @xmath16 , and @xmath148 . _
again , if @xmath35 fixes @xmath12 , then @xmath35 fixes @xmath16 as well and this contradicts our choice of @xmath31 .
since @xmath35 switches @xmath12 and @xmath16 , @xmath81 and @xmath83 are isomorphic .
in particular , @xmath149 . this proves ( 2 ) .
\(3 ) _ @xmath86 does not fix @xmath16 , and @xmath12 is the unique center of @xmath87 .
_ by the same argument as in the proof of ( 1 ) , @xmath12 is a center of @xmath87 , and if there is another one , then it must be @xmath16 .
suppose @xmath86 fixes @xmath16 .
then , again @xmath12 is fixed as well and this contradicts the definition of @xmath32 by the same argument as in ( 6 ) in the proof of [ aft1 ] .
therefore @xmath86 does not fix @xmath16 .
now , suppose @xmath16 is a center of @xmath87 .
since @xmath86 does not fix @xmath16 , it switches @xmath12 and @xmath16 and @xmath92 is isomorphic to @xmath80 .
but then , @xmath150 , a contradiction .
this proves ( 3 ) . since @xmath12 is the unique center of @xmath87 and @xmath86
does not fix @xmath16 , the component @xmath92 of @xmath98 is isomorphic to another component @xmath44 of @xmath98 .
note that the union of all components of @xmath110 different from @xmath92 is exactly @xmath118 .
and @xmath44 has size @xmath151 .
this implies that there are exactly three components of @xmath110 , namely @xmath92 , @xmath44 , and the third one with a single vertex .
therefore @xmath12 has a neighbor of degree one , and this implies that @xmath31 is a neighbor of @xmath12 .
\(4 ) @xmath153 and @xmath81 are paths .
it is enough to show that @xmath154 is a path since @xmath155 by @xmath35 .
suppose there exists a vertex of degree at least three in @xmath154 .
choose such a vertex @xmath156 with @xmath157 as small as possible .
then , @xmath127 has degree at least three in @xmath158 as well . to see this ,
first observe that @xmath159 has degree at least three in @xmath81 since @xmath160 and @xmath152 .
therefore @xmath161 has degree at least three in @xmath83 since @xmath122 .
but then , @xmath162 @xmath163 this contradicts our choice of @xmath79 .
this proves ( 4 ) . by ( 4 )
, @xmath0 is a tree with a unique vertex of degree three , namely @xmath12 , and one of the three components of @xmath48 consists of a single vertex , namely @xmath31 , and the other two components @xmath81 and @xmath164 are paths .
let @xmath165 ; then @xmath166 by ( 2 ) .
finally if @xmath167 , then deleting the leaf of @xmath0 in @xmath168 yields another automorphism - free tree , and if @xmath169 , then @xmath170 , and so @xmath0 is not automorphism - free .
therefore @xmath171 , and @xmath0 is isomorphic to @xmath11 .
this proves [ main1 ] . | a finite tree @xmath0 with @xmath1 is called _ automorphism - free _ if there is no non - trivial automorphism of @xmath0 .
let @xmath2 be the poset with the element set of all finite automorphism - free trees ( up to graph isomorphism ) ordered by @xmath3 if @xmath4 can be obtained from @xmath5 by successively deleting one leaf at a time in such a way that each intermediate tree is also automorphism - free . in this paper , we prove that @xmath2 has a unique minimal element .
this result gives an affirmative answer to the question asked by rupinski in @xcite . |
baryogenesis is one of the fundamental questions left unanswered by the standard model ( sm ) of particle physics .
there are many approaches to generating the baryon asymmetry of the universe ( bau ) , some examples of which are electroweak baryogenesis ( ewbg ) @xcite , leptogenesis @xcite and affleck - dine baryogenesis @xcite .
ewbg is an intriguing possibility because it relies only upon weak scale physics and gives rise to possible direct experimental tests , but it can not take place within the sm @xcite given the current lower bounds on the higgs mass @xcite .
ewbg could be realized within the minimally supersymmetric standard model ( mssm ) @xcite , see @xcite for reviews , but it requires a particular corner of the mssm parameter space known as the light stop scenario ( lss ) @xcite . as the name suggests there are in principle directly testable predictions of new light particles that can be discovered at the lhc .
however , as with many searches at the lhc , depending on the exact spectra , particles with copious production cross sections can be missed if a particular signature is not investigated .
the benefit of the lss is that direct production of stops are not the only test of the scenario . in the mssm
the stop sector is crucial for a viable higgs sector due to the needed radiative corrections to the higgs mass .
the stops also contribute to various effective higgs couplings , most significantly to two gluons .
this intertwining of the two sectors means that there are additional tests of ewbg in the mssm , based purely on the properties of the higgs .
recently , both the atlas @xcite and cms @xcite experiments at the large hadron collider ( lhc ) reported intriguing @xmath5 excesses in the diphoton and @xmath6 channels that could be interpreted as early signs of a @xmath0 standard - model - like higgs .
more data is needed to claim discovery , but it is not too early to start thinking about what the implications of such a result might be @xcite .
in particular , there are only two ways that such a heavy higgs could be realized within the mssm : large mixing in the stop sector with relatively light stop masses , or minimal mixing with at least one of the stops being extremely heavy .
this potential higgs mass measurement immediately creates tension with ewbg .
a detailed recent analysis of the lss @xcite showed that a higgs mass of @xmath0 requires the left - handed ( lh ) stop to be heavier than about @xmath7 , since the stop mixing can not be large .
the right - handed ( rh ) stop , on the other hand , gets pushed to a mass of @xmath8 .
this spectrum is very peculiar , especially when trying to imagine a corresponding susy breaking scheme @xcite .
nevertheless , it remains as the _ only realization of ewbg in the mssm_. before the recent higgs mass measurement , ewbg was reconcilable with a fairly wide variety of higgs phenomenologies @xcite .
now that the stop spectrum has been so strongly constrained by the higgs mass measurement it may be possible to rule out ewbg purely by determining the properties of the higgs boson .
this is particularly attractive , since one could imagine many ways to hide a light stop from direct searches ( e.g. decay through a displaced vertex ) .
using higgs data represents a model - independent approach to excluding ewbg in the mssm .
we will show in this paper that the correlations between different higgs decay channels and production modes , in particular those which occur via loops compared to those that occur at tree level , make predictions that are already in tension with the data . by combining the available constraints from lhc higgs searches ,
we show that _ ewbg in the mssm is already excluded at the ( 90 ) 98% confidence level ( cl ) _ in the ( non-)decoupling limit .
we also examine the exclusion without the assumption of a @xmath0 higgs .
the higgs searches are still highly constraining , excluding the entire ewbg parameter space at the 90% cl except for a small window of @xmath1 due to a small excess seen in the atlas @xmath9 search .
this paper is organized as follows . in section [ s.review ]
we provide a very brief review of electroweak baryogenesis , which explains the need for the lss in the mssm .
we then discuss the current status of the lss in section section [ s.lss ] and the particular parameter space within the lss dictated by a higgs of mass @xmath0 . in section [ s.fingerprint ]
we investigate the _ fingerprint of ewbg in the mssm _ , the correlations amongst the different production and decay channels of the higgs that are the signature of the lss . in section [ s.expstatus ] we discuss the available experimental data from lhc higgs searches and combine them to exclude the ewbg parameter space at the 90% cl .
we conclude with section [ s.conclusion ] .
various technical details of the h iggs decay width calculations are included in the appendix .
as is well known , to generate a bau the three sakharov conditions @xcite must be satisfied : @xmath10-violation , @xmath11-violation ( otherwise any @xmath10-producing process is cancelled out by its @xmath11-counterpart ) and departure from thermal equilibrium ( to prevent washout of the accumulated @xmath10-excess ) .
electroweak baryogenesis @xcite ( see also @xcite for reviews ) is a mechanism for producing the bau that relies entirely on weak scale physics to satisfy the sakharov conditions .
the triangle anomaly in the electroweak sector of the standard model leads to non - perturbative sphaleron processes at high temperatures that violate baryon number , complex phases in the higgs - fermion couplings provide the necessary @xmath11-violation , and departure from thermal equilibrium is instigated by the electroweak phase transition .
even though the standard model has all the qualitative ingredients for electroweak baryogenesis , the size of the generated baryon asymmetry falls far short of the required value @xcite .
the phase transition is first order but only weakly so , and there is not enough @xmath11 violation .
this means that electroweak baryogenesis can only work in a theory with additional complex phases , as well as weak - scale particles that interact strongly with the higgs sector to give the necessary additional contributions to its thermal potential .
these conditions can be satisfied in the mssm @xcite : contributions from stops to the thermal potential of the higgs can generate a stronger first order phase transition , and there are many new sources of cp violation available .
the task of computing the generated bau can be approximately factorized , into sectors that are responsible for the first order phase transition , and those directly responsible for creating the baryon asymmetry during that phase transition .
therefore we can examine the constraints or evidence for these sectors independently .
provided that a strong enough first order phase transition occurs , computation of the bau involves a complicated tunneling , quantum transport and hydrodynamics calculation .
@xmath11-violating interactions between the plasma and the space - time varying higgs vev generate chiral currents across the bubble wall , which sphalerons in the unbroken phase convert to baryon asymmetry .
this excess then partially survives in the broken phase , where sphalerons are suppressed , as the bubble expands .
there is a vast literature on this calculation @xcite .
the uncertainties are still order one , and tend to err on the optimistic side @xcite .
that being said , the generation of a sufficient bau seems at least possible within the mssm for some suitably chosen gaugino / higgsino parameters ( @xmath12 ) , _ if _ there is a strong enough first order phase transition .
the only severe constraint from this step of the calculation comes from edms @xcite that can arise as a result of the required @xmath11-violating phases .
the required phases are @xmath13 and @xmath14 , where @xmath15 is the higgs sector soft mass .
one - loop edm contributions can be suppressed by making the first and second generation sfermions heavier than @xmath16 , but two - loop contributions involving the chargino and higgs fields are sizable unless @xmath17 ( see e.g. @xcite ) .
this generic bound can be loosened if @xmath18 is strongly suppressed relative to @xmath19 , since the phase of the bino - mass by itself does not generate strong two - loop edm contributions . in this _ bino - driven _
scenario @xcite @xmath20 can take on smaller values .
calculating the strength of the first order phase transition is somewhat more straightforward , and ultimately more constraining , than the baryon density calculation .
a sufficiently strong phase transition requires @xmath21 ( see e.g. @xcite ) , where @xmath22 is the _ critical temperature _ at which the electroweak symmetry breaking vacuum @xmath23 is degenerate with the symmetric minimum @xmath24 . in the one - loop thermal higgs potential
one finds that @xmath25 ( cubic coefficient)/(quartic coefficient ) .
the cubic term comes solely from the thermal contribution and has the form @xmath26 , where @xmath27 is the field dependent thermal mass of the additional scalars in the mssm . to maximize
the strength of the phase transition clearly requires maximizing the new contributions to the cubic term .
given the form of the contribution from the scalars of the mssm , the largest potential contribution will come from the stop sector .
the higgs dependent masses of the stops are given by @xmath28 where @xmath29 .
working along the direction of the zero - temperature higgs - vev in the higgs potential is large , and sufficient for our purposes of demonstrating the effect of stops on @xmath30 . ] , @xmath31 , their mass eigenvalues are @xmath32 ^ 2 . }
\ ] ] in the cubic term of the thermal one - loop higgs potential one has to replace the soft masses @xmath33 by @xmath34 , where @xmath35 , @xmath36 are the thermal masses of the lh and rh stops .
( this is necessary to control ir divergences in the one - loop thermal potential and restore the validity of the perturbative expansion at the critical temperature @xcite ) . to maximize the cubic higgs term , one of the stop mass eigenvalues should therefore be close to @xmath37 .
this requires small stop mixing , as well as a fairly precise cancellation between the light stop s thermal mass , @xmath38 , and the necessarily negative stop soft mass - squared @xcite .
this yields one stop that is lighter than the top . to increase the higgs mass beyond the lep limit @xcite and avoid large corrections to the @xmath39-parameter @xcite the lh stop
should then be heavier than a tev or so .
two - loop corrections @xcite are quite large because @xmath40 enters for the first time at this order .
both two - loop and non - perturbative corrections @xcite enhance the phase transition , enlarging the viable parameter space .
this provides a more complete picture of the viable regions of mssm parameter space for electroweak baryogenesis , but the intuition from examining one - loop effects still provides a helpful guide . putting all these ingredients together leads to the _ light stop scenario _ ( lss)@xcite , the only corner of mssm parameter space where electroweak baryogenesis might be possible .
the constraints on the stop sector parameters are the following : * achieving a strong phase transition and avoiding color - breaking requires a mostly right - handed light stop with @xmath41 and @xmath42 .
the mostly left - handed stop should be heavier than @xmath43 to satisfy the lep higgs mass bound ( for a sm - like higgs ) and avoid large corrections to the @xmath39-parameter .
* the gluino should be heavier than @xmath44 to decouple it from the plasma , otherwise its large contribution to the stop thermal masses would make it even more difficult to achieve the needed cancellation @xmath45 .
in addition , there are some constraints on the electroweak gaugino and higgsino parameters to allow for sufficient generation of bau : * @xmath46 or @xmath47 with sufficiently large @xmath11-violating phases in the -ino sector , as well as @xmath48 @xcite . *
@xmath17 , unless all the @xmath11-violation is pushed into the bino soft mass @xcite .
it is well known to experts that the possible detection of a @xmath49 gev higgs at the lhc spells trouble for electroweak baryogenesis in the mssm .
such a large higgs mass can only be achieved if stop mixing is large ( incompatible with a strong phase transition ) or if the left - handed stop is extraordinarily heavy . to quantify the exact consequences of such a heavy higgs
, we draw upon the results of @xcite .
the authors of @xcite studied the electroweak baryogenesis window of the mssm in great detail .
they constructed a low - energy effective theory @xcite in which all scalar superpartners with the exception of the rh stop are pushed to some high common scale .
this effective description , tailored to the light stop scenario , included the most important one- and two - loop effects .
they constructed the resulting thermal higgs potential and scanned over the stop- and higgs - sector parameter space . requiring a sufficiently strong first - order phase transition and avoiding color - breaking yields regions of the stop - higgs mass plane where electroweak baryogenesis could proceed within the mssm .
as expected , a higgs mass in the range of @xmath50 is extremely difficult to accommodate .
the stop sector has to take on a very particular form : @xmath51 with stop mixing being completely negligible for such large @xmath52 .
the size of the allowed @xmath53 range is somewhat overestimated , since it was obtained by interpreting the results of the analysis in a _ very _ conservative fashion . therefore , if this stop spectrum can be excluded , then electroweak baryogenesis in the mssm is excluded ( assuming of course that the higgs mass falls into the above mentioned range ) .
the extremely heavy left - handed stop is in significant conflict with notions of naturalness , reminiscent of split supersymmetry @xcite .
one could ask how a high - energy theory of susy breaking could generate such a spectrum @xcite , but let us put aside such considerations and focus on the phenomenology . the light right - handed stop with a mass of @xmath54
is an extremely interesting _ prediction _ of electroweak baryogenesis within the mssm , emerging as a direct consequence of requiring a sufficiently strong electroweak phase transition and a higgs mass of @xmath0 .
it is already excluded if it decays promptly @xcite or escapes the detector @xcite , but one could imagine it being hidden from direct stop searches somehow , for example by decaying via a displaced vertex @xcite .
the question is then : given a higgs mass of @xmath0 , can electroweak baryogenesis within the mssm be excluded in a model - independent way ?
as it turns out , the answer is yes .
the specific spectrum required by the lss in light of such a higgs mass , especially the light rh stop , makes very definite predictions for the higgs production rate and branching ratios .
this allows us to test electroweak baryogenesis within the mssm using pure higgs data , separate from collider searches for the stop and questions of how such a strange spectrum could be generated by a high - energy theory of susy - breaking .
the presence of a light rh stop can significantly alter higgs production and decay rates compared to their sm expectation . in the context of overall inclusive production cross sections
this has been investigated in detail by @xcite . however , even without an unambiguous @xmath55 higgs discovery at the lhc , or an extremely precise measurement of the @xmath9 branching fraction , it is _ still _ possible to conclusively test the mechanism of ewbg in the mssm .
this is because the lss makes specific predictions for _ all _ possible production and decay modes of the higgs , and they have very particular correlations .
the presence of the light rh stop affects higgs phenomenology through loop level production via gluon fusion and decays to @xmath9 .
the effects are encoded by examining the partial widths , which can be related to both production and decay .
the leading order contributions to gluon fusion ( in the decoupling limit ) are @xcite @xmath56 where @xmath57 , @xmath58 is the normalized higgs coupling to the right - handed stop , which in the lss is given by @xmath59 .
the functions @xmath60 ( @xmath61 or @xmath62 ) are defined as @xmath63/\tau^2\\ a_{1/2}(\tau)&= & 2 \left[\tau+(2\tau-1)f(\tau)\right]/\tau^2\nonumber\\ a_1(\tau)&= & -\left[2\tau^2 + 3\tau+3(2\tau-1)f(\tau)\right]/\tau^2\nonumber\end{aligned}\ ] ] where @xmath64 ^ 2 & \tau > 1 \end{array}\right . .\ ] ] the crucial point is that the light stop loop interferes constructively with the top quark loop , which leads to a more than three - fold _ increase _ in the higgs production cross section via gluon fusion .
however , when investigating the clean @xmath9 decay channel we must also examine the stop s contribution to the @xmath65 decay width , which at lowest order ( again in the decoupling limit ) is @xmath66 where we have also included the contribution from charginos .
we do not explicitly calculate @xmath67 as a function of the chargino mass , but they can shift the width at most by order @xmath68 if their mass is small , @xmath69 gev , and we include this as a theory uncertainty . unlike for the gluon width , the stops destructively interfere with the dominant contribution in eq .
( [ e.hgagadecay ] ) coming from the @xmath70 bosons , and thus _ decrease _ the decay width @xmath67 by nearly a factor of @xmath71 compared to the sm expectation .
thus , while a light stop can effect both @xmath72 and @xmath67 significantly , the effect can be washed out by looking at the total rate of @xmath73 only . however , there are both additional higgs production and decay modes available at the lhc . in particular ,
examining both tree level and loop level ( affected by the stop ) production and decay modes in various combinations should reveal strong correlations amongst the various channels .
we call this the _ fingerprint of electroweak baryogenesis in the mssm_. for example , in gluon fusion events with tree level decays one would expect a large increase over standard model rates . however , if one examined vector boson fusion production of a higgs which then decayed to @xmath9 , we would expect a rate smaller than the sm . for ewbg we can give predictions for all the channels , and even without a higgs discovery one can still make strong exclusion statements by considering their correlations .
this is similar in spirit to analyses investigating naturalness and general composite higgs sectors @xcite .
we use @xcite to compute the ewbg predictions for the higgs decay widths . to understand the various correlations and predictions for ewbg quantitatively
we define ratios of the various production channels , gluon fusion ( ggf ) , vector boson fusion ( vbf ) and associated production ( ap ) , in the lss compared to the sm : @xmath74 @xmath75 is derived by taking ratios of decay widths , while @xmath76 in the decoupling limit .
similarly we can define ratios for the branching fractions @xmath77 compared to the sm as : @xmath78 combining these various production and decay channels we can define a _ partial signal strength _
@xmath79 where @xmath80 labels the higgs decay final state and @xmath81 represents the production channel . as an example , our partial signal strengths for @xmath9 final states when produced through gluon fusion and vector boson fusion are @xmath82 in principle these can both be measured separately , as discussed in section [ s.expstatus ] , in which case a large discernible difference compared to the sm should be found if ewbg takes place within the mssm .
we also consider searches sensitive to both ggf and vbf , with possibly different efficiencies @xmath83 for each production mode . in this case case
the @xmath84 signal strength prediction ( similarly for other channels ) is given by @xmath85 and @xmath86 if the efficiencies for ggf and vbf are comparable . while ewbg fixes the stops of the mssm to particular values , in principle the value of @xmath20 can range from the decoupling limit to very low values of @xmath20 as in bino - driven ewbg@xcite , while still preserving the successes of ewbg .
the value of @xmath20 is important since it alters the @xmath87 coupling compared to the sm , which in turn rescales @xmath88 where @xmath89 is the effective @xmath11-even higgs mixing angle .
( in the decoupling limit , @xmath90 thus , @xmath76 in this case . )
our analysis will take this allowed range for @xmath20 into account .
for technical details on the decay width and cross section ratio computation , as well as the associated theoretical errors , the reader is referred to the appendix .
lcc @xmath91 & & and 300 gev , for a range of stop masses from 80 - 115 gev including theory errors . shown
are signal strength predictions for each channel , with subscripts indicating an exclusive production mode .
the exception is @xmath92}$ ] , which denotes the signal strength prediction for a @xmath93 search with vbf cuts @xcite , such that @xmath94 in eq .
( [ e.muinclusive ] ) .
the purple line is the sm expectation .
this fingerprint is for @xmath95 , but the dependence on @xmath96 is very small in the @xmath97 neighborhood of higgs masses .
solid red bands indicate the range of predictions for @xmath98 .
the light red bands indicate the theory error at the top and bottom of the stop mass range .
@xmath99 was allowed to vary in the range @xmath100 , but its effect is very small since @xmath96 was taken as a low - energy input .
the rate of decays that are dominated by gluon fusion increases for lighter stop masses , while @xmath84 and channels sensitive to vector boson fusion and associated production are much less affected .
, title="fig:",width=529 ] + + @xmath101 & & and 300 gev , for a range of stop masses from 80 - 115 gev including theory errors . shown
are signal strength predictions for each channel , with subscripts indicating an exclusive production mode .
the exception is @xmath92}$ ] , which denotes the signal strength prediction for a @xmath93 search with vbf cuts @xcite , such that @xmath94 in eq .
( [ e.muinclusive ] ) .
the purple line is the sm expectation .
this fingerprint is for @xmath95 , but the dependence on @xmath96 is very small in the @xmath97 neighborhood of higgs masses .
solid red bands indicate the range of predictions for @xmath98 .
the light red bands indicate the theory error at the top and bottom of the stop mass range .
@xmath99 was allowed to vary in the range @xmath100 , but its effect is very small since @xmath96 was taken as a low - energy input .
the rate of decays that are dominated by gluon fusion increases for lighter stop masses , while @xmath84 and channels sensitive to vector boson fusion and associated production are much less affected .
, title="fig:",width=529 ] combining the shifts for @xmath102 and @xmath75 from ( [ e.hggdecay ] ) and ( [ e.hgagadecay ] ) with all the other channels allows us to map the theoretical fingerprint of the entire ewbg scenario over the allowed range of stop masses .
we show this fingerprint in fig .
[ f.fingerprint ] for the decoupling limit and for @xmath103 .
the decoupling limit is required for generic @xmath11-violating phases , and the latter value of @xmath20 was chosen because , as we will see in the next section , it _ minimizes _ the tension between experimental data and the ewbg prediction .
what is striking about fig .
[ f.fingerprint ] is that even including a range of stop masses , theory uncertainties , and @xmath20 , there are easily discernible correlations amongst the various channels .
while not all of these channels have been measured to a very precise level , the particular fingerprint of ewbg in the mssm means that this scenario can in principle be ruled out by combining information from the 7 tev lhc higgs searches only .
of course , it is possible that the higgs may not ultimately be found to have a mass of @xmath0 .
however , because the deviations from sm are so large for ewbg , it is still possible to bound ewbg for arbitrary higgs mass given the current data sets .
we will examine this in detail in the following sections , but the fingerprint detailed in fig .
[ f.fingerprint ] stresses the power of setting limits even with channels that may not have enough data for a discovery with the entire 2012 8 tev data set .
eventually , the study of correlations such as these could be one of our most powerful tools into discerning the effects of new physics , if it is related to electroweak symmetry breaking .
we will start by briefly outlining the available experimental data before moving on to show the extent to which different regions of ewbg parameter space are excluded , both with and without the assumption of a @xmath0 higgs .
cc|c|c|c|c||c|c| & & & + & & ggf & vbf & ap & inclusive & source & @xmath96 range ( gev ) + & & & & & @xmath104 & official & ( 110 , 150 ) + & & @xmath104 & @xmath104 & & & reconstructed@xmath105 @xcite & ( 120 , 128 ) + & & & & & @xmath104 & official & ( 110 , 150 ) + & & & & & @xmath104 & reconstructed@xmath105 @xcite & ( 120 , 128 ) + & & & & & @xmath104 & official & ( 110 , 150 ) + & & @xmath106 & @xmath106 & & & & + & & & & @xmath104 & & official & ( 110 , 130 ) + & & & & @xmath104 & & reconstructed@xmath105 @xcite & ( 110 , 130 ) + & & & & @xmath104 & & official & ( 100 , 150 ) + & & & @xmath106 & & @xmath104 & reconstructed @xcite & ( 110 , 150 ) + & & @xmath106 & @xmath106 & & & & ( 110 , 150 ) + lcc & @xmath95 & @xmath107 + @xmath108 & .
subscripts indicate exclusive production via a single mode . for each channel
we show up to three bands : the ewbg prediction , with @xmath98 ( red , with theoretical error bands in light red ) and the atlas / cms 1-@xmath109 best - fit measurements ( blue / green , with central value indicated in dark blue / green ) . in the @xmath110}$ ]
channel we also show the combined tevatron constraint as a fourth band ( purple ) .
the sm prediction is indicated with a horizontal line at @xmath111 .
, title="fig:",width=321 ] & .
subscripts indicate exclusive production via a single mode . for each channel
we show up to three bands : the ewbg prediction , with @xmath98 ( red , with theoretical error bands in light red ) and the atlas / cms 1-@xmath109 best - fit measurements ( blue / green , with central value indicated in dark blue / green ) . in the @xmath110}$ ]
channel we also show the combined tevatron constraint as a fourth band ( purple ) .
the sm prediction is indicated with a horizontal line at @xmath111 . ,
title="fig:",width=321 ] + + @xmath112 & .
subscripts indicate exclusive production via a single mode . for each channel
we show up to three bands : the ewbg prediction , with @xmath98 ( red , with theoretical error bands in light red ) and the atlas / cms 1-@xmath109 best - fit measurements ( blue / green , with central value indicated in dark blue / green ) . in the @xmath110}$ ]
channel we also show the combined tevatron constraint as a fourth band ( purple ) .
the sm prediction is indicated with a horizontal line at @xmath111 .
, title="fig:",width=321 ] & .
subscripts indicate exclusive production via a single mode . for each channel
we show up to three bands : the ewbg prediction , with @xmath98 ( red , with theoretical error bands in light red ) and the atlas / cms 1-@xmath109 best - fit measurements ( blue / green , with central value indicated in dark blue / green ) . in the @xmath110}$ ]
channel we also show the combined tevatron constraint as a fourth band ( purple ) .
the sm prediction is indicated with a horizontal line at @xmath111 .
, title="fig:",width=321 ] table [ t.channels ] summarizes all the available higgs searches to date that are relevant to our analysis .
a few remarks are in order : * we use eqns .
( [ e.partialsigstrength2 ] ) and ( [ e.muinclusive ] ) to compute the theory predictions for the @xmath84 signal strengths ( similarly for the other channels ) .
the inclusive signal strength prediction assumes equal signal efficiencies for ggf and vbf , which is a conservative choice for setting limits .
production is dominated by ggf , but assuming the vbf efficiency to be zero would lead us to slightly overestimate the theory prediction for the signal strength @xmath113 , since ggf is enhanced in our mssm scenario . as we will see , this would increase tension with the data .
therefore , we set the two efficiencies to be equal , while noting that some deviation from this assumption will not invalidate the analysis since @xmath114 is small for this search .
for the @xmath93 search with vbf cuts we set @xmath115 in eq .
( [ e.muinclusive ] ) @xcite . *
official signal strength bounds were not always available for each channel . fortunately , the authors of @xcite reconstructed approximate signal strength likelihoods for the cms @xmath84 , @xmath116 searches by using the information that is publicly available and generating their own event samples . for other searches we used the methods of @xcite , very similar to the ideas of @xcite , to reconstruct approximate likelihoods where necessary .
* we used the older atlas @xmath117 search using @xmath118 of data @xcite rather than the updated version with @xmath119 @xcite .
the latter is significantly more constraining and looks to increase the tension with the ewbg prediction , but there is not enough information available to reliably disentangle the ggf and vbf contributions . the cms @xmath120 search @xcite is omitted because signal strength bound are only reconstructible for @xmath121 .
[ f.resultplot ] compares the signal strength predictions to the experimental signal strength bounds in all available channels , for @xmath122 and @xmath123 .
the results are displayed for these two higgs masses since they are preferred by the cms and atlas @xmath84 searches , respectively .
a visual inspection already reveals some tension with the data .
we will now make this impression quantitative .
given the large error bars in the early higgs data it is perhaps surprising that we can make relatively strong statements regarding the exclusion of electroweak baryogenesis .
this is due to the correlations of the signal strength predictions in the various channels and their dependence on ewbg parameters .
the higgs signal in the various channels depends only very weakly on @xmath99 , since our parameterization takes @xmath96 as a low - energy input and @xmath99 can not be large for successful ewbg .
therefore , for a given higgs mass , the parameter space of ewbg in the mssm is the @xmath124 plane .
once the higgs mass is determined this will be the relevant parameter space to exclude .
after taking into account theory error and the small amount of @xmath99 dependence , each point in the @xmath124 plane maps to a range of signal strength vectors @xmath125 , which constitute the range of experimental predictions for this parameter point . maximizing the signal strength likelihood function @xmath126 , which is obtained from experimental data , over the range of allowed signal strength vectors gives the exclusion for this point in the @xmath124 plane .
unfortunately the signal strength likelihood function @xmath127 is not directly available .
however , we can obtain a passable approximation by first assuming that the separate searches are independent , and then using the 1-@xmath109 best - fit bounds on the separate signal strengths to obtain gaussian approximations for @xmath128 ( taking into account asymmetric error bars where appropriate ) .
normalizing @xmath129 to have a maximum value of zero , we obtain the desired likelihood function .
, obtained by combining the signal strength bounds from the various atlas and cms higgs searches ( not tevatron ) as outlined in section [ ss.availabledata ] . the smallest exclusion at @xmath130 , @xmath131 is 97.2% , which increases to 98.5% if we enforce the decoupling limit ( @xmath132 ) .
, width=340 ] in fig .
[ f.exclusionplot ] we show the exclusion across ewbg parameter space , obtained by combining atlas and cms data for @xmath95 .
the entire parameter space is excluded at the 97.2 % cl ( 98.5 % if we enforce the decoupling limit ) .
the least excluded points are at @xmath130 , with @xmath53 as high as possible .
one could consider several variations on this plot , each with similar results .
* until the higgs mass is precisely known one should generally consider the exclusion of ewbg over a range of @xmath96 .
this produces a very similar pattern of exclusion to fig .
[ f.exclusionplot ] , with tension minimized for @xmath130 and high @xmath53 .
the exclusion as a function of higgs mass ( from atlas and cms data only ) is : + [ cols="<,^,^,^,^,^,^",options="header " , ] [ ss.generallssexclusion ] one could loosen the assumptions of our analysis , and ask what the available lhc data tells us about a wider range of higgs and stop masses .
dropping the assumption of a 123 - 128 gev higgs allows us to examine the prospects of electroweak baryogenesis in the mssm if the higgs were to sit at a different mass .
[ f.mhmstopexclusion ] shows the exclusion from atlas and cms data as a function of the @xmath133 plane .
this exclusion plot was created via the same method as fig .
[ f.exclusionplot ] , using gaussian approximations of the signal strength bounds . for each point in the @xmath133 plane
we minimize exclusion with respect to theory error , @xmath99 dependence and @xmath20 dependence , using the experimental signal strength bounds for whatever higgs masses they are available ( see table [ t.channels ] ) .
however , there is one additional complication with this expanded higgs mass range : the atlas @xmath134 bounds have extremely asymmetric error bars for @xmath135 .
this suggests a reduced reliability of the gaussian likelihood approximation , and therefore we do not use the atlas @xmath134 bounds for @xmath135 . what does fig .
[ f.mhmstopexclusion ] imply for mssm ewbg in general ? without a higgs mass constraint , the successful electroweak phase transition requires @xmath136 and @xmath137 @xcite .
as we can see , lhc data already excludes almost all of this parameter space at more than @xmath138 cl , with the most notable exception being a ( 117 - 119 ) gev higgs mass window that is only excluded at @xmath139 cl .
including tevatron data does not significantly enhance these constraints . if we were to include the atlas @xmath134 constraints below @xmath140 by naively applying its @xmath141 error bar to @xmath142 and using the gaussian likelihood approximation , then the exclusion of the allowed higgs mass window increases by @xmath143 .
however , given the possibly long tail of the likelihood distribution these numbers must be taken with a grain of salt . at any rate , it is clear that the lhc higgs searches are already excluding large portions of the lss parameter space , even without the assumption of a particular higgs mass .
further data should allow exclusion of the 117 - 119 gev higgs mass window , but these considerations are of course irrelevant if a higgs mass of @xmath0 is confirmed .
we also point out that our exclusions could likely be improved if the experimental collaborations provided more detailed signal strength likelihoods .
finally , we point out that the discovery or exclusion of the mssm pseudoscalar higgs would of course significantly enhance the power of these constraints by restricting the allowed value of @xmath20 .
ewbg in the mssm has long been an attractive possibility for baryogenesis due to the presence of additional light supersymmetric states within reach of colliders . with the 7 tev run of the lhc
now completed , there finally is enough high energy data to confront the entire parameter space of ewbg .
however , in direct searches there are in principle ways to weaken constraints without affecting the physics of ewbg . in this paper
we looked at the indirect consequences for the higgs sector alone , which represents an irreducible constraint on ewbg in the mssm .
given that the only window for ewbg that still exists is the lss , there are very large corrections to higgs phenomenology and in particular strong correlations amongst channels as we have demonstrated .
we have shown that in the context of a possible higgs mass measurement of @xmath144 at the lhc , ewbg in the mssm is now ruled out at greater than a 98% cl in the decoupling limit for the higgs sector , and at least 90% cl for lighter values of @xmath20 .
this is primarily due to the fact that this heavy of higgs mass even further restricts the allowed parameter space for ewbg in the mssm .
if we relax the constraint of @xmath144 , to allow for a larger parameter space , we can _ still _ constrain the lss in full generality with the current lhc data .
we find that the only significant region excluded at less than @xmath145 cl is @xmath146 - @xmath147 .
this window can be excluded with more data , though of course the point is moot if the higgs is confirmed to have a mass of @xmath0 .
finally , one can speculate that lhc higgs searches could in fact offer an almost model - independent way to exclude electroweak baryogenesis , even beyond the mssm implementation . at its very basic level ,
the mechanism of ewbg requires the presence of weak - scale particles with substantial higgs couplings to generate the strong first - order phase transition via their contributions to the thermal potential .
these particles , via those very same couplings , could then significantly contribute to the effective higgs couplings and allows for testing ewbg in other scenarios @xcite .
while complete exclusion of the electroweak baryogenesis idea may not be possible @xcite , this could nonetheless reduce the array of viable models to a few very special cases .
ultimately , the indirect tests of higgs phenomenology using correlations , such as those we have demonstrated in this paper , may prove to be the most important window into new physics .
we would like to thank sally dawson for many helpful conversations .
we are also very grateful to tomer volansky and eric kuflik for supplying us with their reconstructed cms signal strength likelihoods from @xcite , as well as useful discussions .
the work of d.c . was supported in part by the national science foundation under grant phy-0969739 .
the work of p.j . was supported in part by the u.s .
national science foundation under grant nsf - phy-0969510 , the lhc theory initiative , jonathan bagger , pi .
the work of p.m. was supported in part by nsf career award nsf - phy-1056833 .
most of the public programs available for branching ratio and higgs production / decay calculations of the mssm higgs bosons are unsuitable for the unusual stop spectrum under consideration .
we have therefore implemented a simple ` mathematica ` code to perform the computations following the references @xcite .
this is feasible since we only care about calculating decay widths and production cross sections from a very simple low - energy spectrum , not the derivation of that spectrum from uv - parameters .
the case of small @xmath20 is handled as follows : for each choice of @xmath148 and @xmath20 , we use the expressions in @xcite to derive the lh stop mass required to give the desired higgs mass at one - loop resummed order .
this allows us to compute the radiatively corrected charged and heavy neutral higgs masses at the same order and include their contributions in the light higgs decay widths and branching fractions .
this one - loop resummed calculation might be insufficient if we were actually interested in how exactly the higgs spectrum is derived from a low - energy theory , but for the purposes of ` sweeping through the higgs spectrum ' as we change @xmath20 this is certainly expected to work well .
the branching ratio calculations in our code include most of the important higher order corrections under the assumption of no stop mixing .
understanding the theoretical uncertainties is quite important for a comparison with lhc data .
therefore we briefly summarize the decay width calculation for each channel and estimate the theory error from those higher order corrections we neglected , as well as chargino and neutralino contributions . *
@xmath149 : the lo contribution is a one - loop diagram with quarks or squarks running in the loop . only the light right - handed stops were considered in our calculation since the contribution from the heavy scalars decouples .
most of the important nlo / nnlo qcd corrections @xcite were implemented with the exception of those from gluino loops . however , the latter are quite small ( a few percent for @xmath150 tev ) @xcite
. the electroweak ( ew ) corrections @xcite are @xmath151 due to large cancellations from various contributions . *
@xmath152 : this channel is also loop induced at lo .
however , in addition to quarks and squarks , there is also a small contribution from charginos and charged higgs in the loop . while the contributions are small for @xmath153 gev , they can be as big as @xmath154 for @xmath69 gev @xcite .
the contributions from the charged higgs are ignored since they are assumed to be heavy and therefore decouple .
most of the nlo / nnlo qcd corrections @xcite are implemented , an exception being the qcd corrections to the squark loop diagrams leading to an error of @xmath155 @xcite .
the ew corrections from the top yukawa coupling @xcite is also implemented for the @xmath65 channel . *
@xmath156 : the vector boson @xmath157 here refers to the massive gauge bosons @xmath70 and @xmath158 . since , we assume that the higgs mass is @xmath159 gev , one of the final state vector bosons is always off - shell @xcite .
the nlo ew corrections @xcite have been implemented for this channel and we expect the remaining theoretical uncertainties to be small . in the decoupling limit the @xmath160 couplings are sm - like , so the partial decay width is identical to the sm case . for small @xmath20
the tree - level coupling is rescaled ( see below ) .
the nlo susy contributions have been ignored as they are expected to be small .
* @xmath161 : for a @xmath162 gev higgs , the decay to top quarks is kinematically forbidden and thus the relevant channels are decays to bottom , charm , and tau . we have implemented nlo / nnlo qcd corrections @xcite including the running of the quark masses to absorb the large logarithms @xcite .
nlo ew contributions have also been implemented @xcite . in the mssm ,
the bottom quark yukawa coupling gets significant corrections which are non - vanishing in the decoupling limit @xcite . in our scenario ,
the corrections to the bottom quark yukawa are given by @xmath163 .
since the left - handed sbottom is extremely heavy we conclude that these susy qcd corrections are small and can be ignored .
the only significant source of error in this channel is the uncertainty in the quark masses since they can not be directly measured .
this translates to an error of @xmath164 @xmath165 for the channel @xmath166 @xcite .
the lo cross - section for a @xmath167 process is proportional to the decay width of the inverse process .
therefore we can use the following approximation to estimate the mssm higgs production cross - section through gluon fusion : @xmath168 where we use the decay widths calculated above .
the qcd k - factors for @xmath169 differ by @xmath170 @xmath165 for the sm and the mssm case and this difference is taken into account by using nlo decay widths . thus we expect this approximation to work very well , and take its contribution to the theory error of our signal strength prediction to be small ( compared to the other sources of uncertainty ) , of order a few percent .
the vector boson fusion and associated production cross section ratios are @xmath171 in the decoupling limit , but for small @xmath20 the @xmath87 tree - level couplings are rescaled compared to the sm
. therefore @xmath88 where @xmath89 is the effective @xmath11-even higgs mixing angle .
( in the decoupling limit , @xmath90 thus , @xmath76 in this case . )
we expect the error introduced by this approximation to also be a few percent
. finally , to compute theory predictions for inclusive signal strengths we need @xmath172 the sm higgs production cross section and associated theoretical errors are calculated in @xcite , which we use to obtain predictions and theoretical uncertainties for @xmath114 .
v. a. kuzmin , v. a. rubakov and m. e. shaposhnikov , phys . lett .
b * 155 * , 36 ( 1985 ) ; f. r. klinkhamer and n. s. manton , phys .
d * 30 * , 2212 ( 1984 ) ; p. b. arnold and l. d. mclerran , phys .
d * 36 * , 581 ( 1987 ) ; p. b. arnold and l. d. mclerran , phys .
d * 37 * , 1020 ( 1988 ) ; s. y. .khlebnikov and m. e. shaposhnikov , nucl .
b * 308 * , 885 ( 1988 ) .
j. m. cline , hep - ph/0609145 ; m. trodden , rev .
phys . * 71 * , 1463 ( 1999 ) [ hep - ph/9803479 ] ; a. riotto , hep - ph/9807454 ; a. riotto and m. trodden , ann .
nucl . part .
* 49 * , 35 ( 1999 ) [ hep - ph/9901362 ] ; m. quiros , hep - ph/9901312 .
g. d. moore , jhep * 0003 * , 006 ( 2000 ) [ hep - ph/0001274 ] ; p. john and m. g. schmidt , nucl .
b * 598 * , 291 ( 2001 ) [ erratum - ibid .
b * 648 * , 449 ( 2003 ) ] [ hep - ph/0002050 ] ; p. john and m. g. schmidt , hep - ph/0012077 ; a. megevand and a. d. sanchez , nucl .
b * 825 * , 151 ( 2010 ) [ arxiv:0908.3663 [ hep - ph ] ] ; j. m. moreno , m. quiros and m. seco , nucl .
b * 526 * , 489 ( 1998 ) [ hep - ph/9801272 ] .
v. cirigliano , m. j. ramsey - musolf , s. tulin and c. lee , phys .
d * 73 * , 115009 ( 2006 ) [ hep - ph/0603058 ] .
d. j. h. chung , b. garbrecht , m. j. ramsey - musolf and s. tulin , phys .
lett . * 102 * , 061301 ( 2009 ) [ arxiv:0808.1144 [ hep - ph ] ] .
g. -c .
cho and k. hagiwara , nucl .
b * 574 * , 623 ( 2000 ) [ hep - ph/9912260 ] ; s. heinemeyer , w. hollik and g. weiglein , phys .
rept . * 425 * , 265 ( 2006 ) [ hep - ph/0412214 ] ; s. dawson and p. jaiswal , phys .
d * 82 * , 073017 ( 2010 ) [ arxiv:1009.1099 [ hep - ph ] ] . n. arkani - hamed and s. dimopoulos , jhep * 0506 * , 073 ( 2005 ) [ hep - th/0405159 ] ; g. f. giudice and a. romanino , nucl .
b * 699 * , 65 ( 2004 ) [ erratum - ibid .
b * 706 * , 65 ( 2005 ) ] [ hep - ph/0406088 ] ; n. arkani - hamed , s. dimopoulos , g. f. giudice and a. romanino , nucl .
b * 709 * , 3 ( 2005 ) [ hep - ph/0409232 ] , v. m. abazov _ et al .
_ [ d0 collaboration ] , phys .
b * 696 * , 321 ( 2011 ) [ arxiv:1009.5950 [ hep - ex ] ] .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
lett . * 104 * , 251801 ( 2010 ) [ arxiv:0912.1308 [ hep - ex ] ] .
y. kats and d. shih , jhep * 1108 * , 049 ( 2011 ) [ arxiv:1106.0030 [ hep - ph ] ] .
g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 701 * , 1 ( 2011 ) [ arxiv:1103.1984 [ hep - ex ] ] ; g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 703 * , 428 ( 2011 ) [ arxiv:1106.4495 [ hep - ex ] ] ; v. m. abazov _ et al .
_ [ d0 collaboration ] , [ arxiv:1110.3302 [ hep - ex ] ] . k. g. chetyrkin , b. a. kniehl and m. steinhauser , phys . rev .
lett . * 79 * , 353 ( 1997 ) [ hep - ph/9705240 ] .
m. spira , a. djouadi , d. graudenz and p. m. zerwas , nucl .
b * 453 * , 17 ( 1995 ) [ hep - ph/9504378 ] .
s. dawson and r. kauffman , phys .
d * 49 * , 2298 ( 1994 ) [ hep - ph/9310281 ] .
h. - q. zheng and d. - d. wu , phys .
d * 42 * , 3760 ( 1990 ) ; a. djouadi , m. spira , j. j. van der bij and p. m. zerwas , phys .
b * 257 * , 187 ( 1991 ) ; s. dawson and r. p. kauffman , phys .
d * 47 * , 1264 ( 1993 ) ; k. melnikov and o. i. yakovlev , phys .
b * 312 * , 179 ( 1993 ) [ hep - ph/9302281 ] .
j. fleischer , o. v. tarasov and v. o. tarasov , phys .
b * 584 * , 294 ( 2004 ) [ hep - ph/0401090 ] .
m. s. chanowitz , m. a. furman and i. hinchliffe , phys .
b * 78 * , 285 ( 1978 ) ; j. fleischer and f. jegerlehner , phys .
d * 23 * , 2001 ( 1981 ) ; z. hioki , phys .
b * 224 * , 417 ( 1989 ) [ erratum - ibid .
b * 228 * , 560 ( 1989 ) ] .
s. g. gorishnii , a. l. kataev , s. a. larin and l. r. surguladze , mod .
lett . a * 5 * , 2703 ( 1990 ) ; k. g. chetyrkin and m. steinhauser , phys .
b * 408 * , 320 ( 1997 ) [ hep - ph/9706462 ] ; k. g. chetyrkin and a. kwiatkowski , nucl .
b * 461 * , 3 ( 1996 ) [ hep - ph/9505358 ] ; s. a. larin , t. van ritbergen and j. a. m. vermaseren , phys .
b * 362 * , 134 ( 1995 ) [ hep - ph/9506465 ] .
e. braaten and j. p. leveille , phys .
d * 22 * , 715 ( 1980 ) ; n. sakai , phys .
d * 22 * , 2220 ( 1980 ) ; t. inami and t. kubota , nucl .
b * 179 * , 171 ( 1981 ) ; s. g. gorishnii , a. l. kataev and s. a. larin , sov .
j. nucl .
phys . *
40 * , 329 ( 1984 ) [ yad .
fiz . * 40 * , 517 ( 1984 ) ] ; m. drees and k. -i .
hikasa , phys .
d * 41 * , 1547 ( 1990 ) .
b. a. kniehl , nucl .
b * 376 * , 3 ( 1992 ) ; d. y. .bardin , b. m. vilensky and p. k. .khristova , sov .
j. nucl .
phys . * 53 * , 152 ( 1991 ) [ yad .
* 53 * , 240 ( 1991 ) ] ; a. dabelstein and w. hollik , z. phys .
c * 53 * , 507 ( 1992 ) ; b. a. kniehl , phys .
d * 50 * , 3314 ( 1994 ) [ hep - ph/9405299 ] ; a. djouadi and p. gambino , phys .
d * 49 * , 4705 ( 1994 ) [ hep - ph/9308338 ] ; b. a. kniehl and m. spira , nucl .
b * 432 * , 39 ( 1994 ) [ hep - ph/9410319 ] ; a. kwiatkowski and m. steinhauser , phys .
b * 338 * , 66 ( 1994 ) [ erratum - ibid .
b * 342 * , 455 ( 1995 ) ] [ hep - ph/9405308 ] .
h. e. haber , m. j. herrero , h. e. logan , s. penaranda , s. rigolin and d. temes , phys .
d * 63 * , 055004 ( 2001 ) [ hep - ph/0007006 ] ; j. guasch , w. hollik and s. penaranda , phys .
b * 515 * , 367 ( 2001 ) [ hep - ph/0106027 ] .
lhc higgs cross section working group , s. dittmaier , c. mariotti , g. passarino , and r. tanaka ( eds . ) , _ handbook of lhc higgs cross sections : 1 .
inclusive observables _ , cern-2011 - 002 ( cern , geneva , 2011 ) , arxiv:1101.0593 [ hep - ph ] . | in the context of the mssm the light stop scenario ( lss ) is the only region of parameter space that allows for successful electroweak baryogenesis ( ewbg ) .
this possibility is very phenomenologically attractive , since it allows for the direct production of light stops and could be tested at the lhc .
the atlas and cms experiments have recently supplied tantalizing hints for a higgs boson with a mass of @xmath0 .
this higgs mass severely restricts the parameter space of the lss , and we discuss the specific predictions made for ewbg in the mssm . combining data from all the available atlas and cms higgs searches reveals a tension with the predictions of ewbg even at this early stage .
this allows us to exclude ewbg in the mssm at greater than ( 90 ) 98% confidence level in the ( non-)decoupling limit , by examining correlations between different higgs decay channels .
we also examine the exclusion without the assumption of a @xmath0 higgs .
the higgs searches are still highly constraining , excluding the entire ewbg parameter space at greater than 90% cl except for a small window of @xmath1 .
2.5 cm 0.2 cm * david curtin@xmath2 , prerit jaiswal@xmath3 , patrick meade@xmath2 * _ @xmath4 c. n. yang institute for theoretical physics + stony brook university , stony brook , ny 11794 .
_ + + [email protected] , [email protected] , [email protected] 0.3truecm |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.